Ice Sheet Melting Throughout Mars Climate History: Mechanisms, Rates, and Implications by Kathleen Elizabeth Scanlon S.B., The University of Chicago, 2008 M.S., The University of Hawai‘i at Mānoa, 2010 Sc.M., Brown University, 2012 A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Earth, Environmental, and Planetary Sciences at Brown University Providence, Rhode Island May 2016 © Copyright 2016 by Kathleen E. Scanlon This dissertation by Kathleen E. Scanlon is accepted in its present form by the Department of Earth, Environmental and Planetary Sciences as satisfying the dissertation requirement for the degree of Doctor of Philosophy. Date James W. Head III, Ph.D., advisor Recommended to the Graduate Council Date Yongsong Huang, Ph.D., reader Date Jung-Eun Lee, Ph.D., reader Date John F. Mustard, Ph.D., reader Date James L. Fastook, Ph.D., reader Approved by the Graduate Council Date Peter M. Weber Dean of the Graduate School iii Kathleen Scanlon kathleen_scanlon@brown.edu Date of birth: December 30, 1985 Place of birth: Cleveland, Ohio, USA Education Brown University, August 2010 – present. Sc.M. in Geological Sciences, 2012. Thesis title: Orographic precipitation in valley network headwaters: Constraints on the ancient Martian atmosphere. Ph.D. in Earth, Environmental and Planetary Sciences, expected May 2016. Dissertation title: Ice Sheet Melting Throughout Mars Climate History: Mechanisms, Rates, and Implications. Advisor: James Head. The University of Hawai'i at Mānoa, August 2008 – December 2010 M.S. in Meteorology, 2010. Thesis title: Eddy-induced poleward migration of westerly jets in an idealized atmospheric global circulation model. Advisor: Fei-Fei Jin. The University of Chicago, September 2004 – June 2008 S.B. in Geophysical Sciences (with Honors), 2008; fulfilled degree requirements for the A.B. in Physics, 2008. Awards and Funding • LPI Career Development Award for travel to LPSC 47, March 2016 • Brown University, Fifth-Year Fellowship, 2014 – 2015 • NASA Astrobiology Institute support to attend the Astrobiology Grand Tour, June 2013 • NASA Graduate Student Researchers Program, 2012 – 2014 • NASA – Nordic Astrobiology Summer School, July 2012 • Brown University, University Fellowship, 2010 – 2011 • University of Chicago, Departmental Honors in Geophysical Sciences, June 2008 • University of Chicago, Jeff Metcalf Fellowship, summer 2006 • University of Chicago, National Merit Scholarship, 2004 – 2008 • NASA Glenn Research Center SHARP Student of the Year, summer 2002 Public Outreach • Taught special, hands-on science lessons for students at Vartan Gregorian Elementary School each semester with fellow Brown Geology graduate students, 2011 – 2015 Peer-Reviewed Publications 1. Scanlon, K. E., J. W. Head, and D. R. Marchant (2015), Remnant Buried Ice in the Equatorial Regions of Mars: Morphological Indicators Associated with the Arsia Mons Tropical Mountain Glacier Deposits, Planetary and Space Science, 111, 144–154. 2. Scanlon, K. E., J. W. Head, and D. R. Marchant (2015), Volcanism-induced, local wet- based glacial conditions recorded in the Late Amazonian Arsia Mons tropical mountain glacier deposits, Icarus, 250, 18–31. 3. Scanlon, K. E., J. W. Head, L. Wilson, and D. R. Marchant (2014), Volcano–ice interactions in the Arsia Mons tropical mountain glacier deposits, Icarus, 237, 315– iv 339. 4. Scanlon, K. E., J. W. Head, J.-B. Madeleine, R. D. Wordsworth, and F. Forget (2013), Orographic precipitation in valley network headwaters: Constraints on the ancient martian atmosphere, Geophysical Research Letters, 40, 4182–4187. 5. Zypman, F. R., J. Ferrante, M. Jansen, K. Scanlon, and P. Abel (2003), Evidence of self-organized criticality in dry sliding friction, Journal of Physics: Condensed Matter, 15, L191–L196. Selected Oral Conference Presentations 1. Scanlon K. E., and J. W. Head. The Dorsa Argentea Formation: Insights into Early Mars Climate Change. Presented at LPSC XLVI, 2015, The Woodlands, Texas, USA. 2. Scanlon K. E., and J. W. Head. Volcano-Ice Interactions at Arsia Mons, Mars. Presented at the Fourth Moscow International Solar System Symposium, 2013, Moscow, Russia. 3. Scanlon K. E., and J. W. Head. Volcano-Ice Interactions at Arsia Mons, Mars. Presented at LPSC XLIV, 2013, The Woodlands, Texas, USA. 4. Scanlon, K. E. (invited). Introduction to Geology for Non-Geologists. Presented at the NASA – Nordic Astrobiology Summer School, 2012, Reykjavík, Iceland. 5. Scanlon, K. E., J. W. Head, J.-B. Madeleine, R. D. Wordsworth, and F. Forget. Orographic Precipitation on Early Mars: Towards New Climate Constraints. Presented at the Third International Conference on Early Mars, 2012, Lake Tahoe, Nevada, USA. 6. Scanlon, K. E., J. W. Head, J.-B. Madeleine, R. D. Wordsworth, and F. Forget., Orographic Precipitation in Terra Cimmeria: Towards New Constraints on the Climate of Noachian Mars. Presented at LPSC XLIII, 2012, The Woodlands, Texas, USA. Recent Employment Experience Brown University Department of Geological Sciences, August 2010 – present Currently, using an energy balance snowmelt model with the LMD Mars Global Climate Model (MGCM) to constrain climates in which the equatorial martian valley networks may have formed; and integrating geomorphology, melt modeling, and climate modeling to elucidate the history of the Dorsa Argentea Formation and hence the Late Noachian – Early Hesperian climate transition on Mars. Previously, modeled scenarios for the emplacement of the martian valley networks and constrained ancient martian atmospheric pressure using the LMD MGCM and an analytical orographic precipitation model; mapped and developed models for the evolution of glaciovolcanic environments in the Arsia Mons fan-shaped deposit; and graded assignments, led a laboratory, and provided one-on-one help to students as teaching assistant for GEOL0050: Earth, Moon and Mars. University of Hawai'i at Mānoa, Department of Meteorology, August 2008 – August 2010 Created software to evaluate a seasonal genesis parameter for tropical cyclone formation in the Pacific under present conditions and in a high-CO2 scenario. Taught laboratory sections and graded lecture sections for MET 101: Introduction to Meteorology. Conducted research in geophysical fluid dynamics using the MIROC atmospheric GCM. v Acknowledgements First: many thanks to my advisor, Jim Head. From suggesting I apply to the Western Australia astrobiology field trip to enthusiastically exchanging ideas about where to go next with the work we started in this thesis, you have gone above and beyond to provide me with opportunities to grow as a scientist. Especially given how busy you are, your thoughtful revisions and rapid turnaround time on our chapter drafts speak volumes to your dedication to your students. The nature of the martian climate at the Noachian-Hesperian boundary was the most interesting problem I could think of when I was applying to Ph.D. programs (and still is); thank you for the opportunity to work on it. Thanks to my earlier research mentors, Mark Jansen, Michelle Bright, Karen Byrum, Ray Pierrehumbert, Kevin Hamilton, and Fei-Fei Jin, for teaching me the research skills and giving me the opportunities that led me here. Thanks to all of my teachers at Brown for equipping me with the knowledge to start a career as a geologist; special thanks to Jim Russell and Amy Barr for providing detailed feedback in their graduate seminars that greatly improved the clarity of my writing and PowerPoint slides, respectively. Special thanks to Jay Dickson for being an invaluable ally in the war against ArcMap, and to Jean-Baptiste Madeleine, Laura Kerber, and Robin Wordsworth for patiently answering my many questions about the GCM. Thanks to Kevin Stacey for getting people outside of academia interested in reading our papers, and to Dave Marchant, Jim Fastook, François Forget, and Lionel Wilson, as well as the co-authors who have already been mentioned, for the constructive criticism that helps make sure the papers are worth reading. Thanks to the NASA Graduate Student Researchers Program vi for financially supporting part of my graduate career, and to the NASA Astrobiology Institute for supporting my travel to two field courses that greatly enriched the way I think about glaciovolcanic environments and ≥ 3.5 billion-year-old rocks, the two primary themes of this thesis. Thanks to Nancy Christy, Janet Eager, Lisa Noble, Melissa DeAugustinis, Pat Davey, and Nancy Fjeldheim for handling so much travel, grant, and thesis paperwork for us grad students, and to Anne Côté for making sure our LPSC travel always goes smoothly and for Figure 31 in Chapter 2. Thanks to all of the other Brown Geology grad students for making my time here so much more fun; my memories of the GCB, LF 105, Loui’s, the Brown hockey rink, The Woodlands, and the seafloor lab are filled with laughter. I feel lucky that I’ll still get to see many of you every year at conferences even when we’ve all graduated; here’s to monopolizing the hotel pool at LPSC when we’re all old as rocks. Thanks also to my former ice hockey teammates on the Rhode Island Panthers and to everyone at Horizon Aviation for reminding me that whether or not there is life outside of Earth, there is life outside of the Lincoln Field Building. Thanks to Aunt Eileen, for always encouraging my interest in art; I wish we could have talked about all the figures I had the opportunity to draw for some of our later papers. Thanks to Aunt Kate and Uncle Dick for teaching me about mindfulness meditation and Bruce Springsteen, both of which have been helpful in finishing this thesis; to Poppa for always being a brief and happy phone call away; to Uncle Jack, John, and James, for making holiday gatherings more fun by your presence; to Grandma Kruse, for your great enthusiasm for both my and “the Greek god’s” endeavors in life; to Aunt Paula and Uncle Mike, for your encouragement in my space goals; to Annie, for always vii checking in on the rest of us even while handling your own dissertation and TA work; to Brian, for the endless GIFs and for your help with the next big career move; and to Patrick, for proudly telling your teachers that your sister was a science major when you were in grade school, and for calling me up sometimes to talk about Genetics or O-Chem now that you’re one too. Thanks to my parents-in-law, Cathy and Harry Poulos, for your emotional support and generosity, to Theresa for being a BodyPump role model, and to Greg for your benevolent rule of Earth-Wilmette. Very special thanks to Nick Poulos for your love and support over the last decade. You encourage me to try challenging but worthwhile things, like applying to planetary geology Ph.D. programs when I still considered myself an atmospheric fluid dynamicist. I thought you were just being a nerd when you said you wanted to start a geology department D&D group, but pizza, beer, and pretending to be a roc-riding Viking ranger with 4 – 8 of the above-mentioned awesome geograds once a week turned out to be one of the most fun parts of grad school; we all appreciate the hours of work you put in every week to run it. Any success I achieve is yours to share, and I’m so proud of the success you’re already finding in your law career. I am very lucky to have you in my life. Finally, this thesis is dedicated to my parents: Patti Scanlon, who kept a scrapbook of the Apollo missions when she was a girl, and Brian Scanlon, who built model kits of Gemini capsules and other NASA spacecraft when he was a boy. Along with everything else you’ve given us, thank you so much for sharing your love of space exploration with your children. viii Preface Branching river valley networks, many connecting voluminous paleolakes, dissect the ancient highlands of Mars, but generally ceased activity near the Noachian – Hesperian boundary ~3.7 billion years ago. The nature of the climate that allowed so much water to flow across the surface of what is now a cold desert planet, and the road by which that climate later evolved to its present state, comprise one of the longstanding mysteries of planetary science. The problem of whether carving the widespread and dendritic valley networks required an Earthlike climate, with oceans and rainstorms, is considered particularly relevant to the problem of whether microbial life ever arose on Mars. Even if there was life on ancient Mars, evidence is rare for long-lived aqueous (let alone habitable) environments at the surface in the most recent ~3 billion years of martian geologic history. In this work, we investigate the melting of snow and ice by intermittent greenhouse warming as an agent for fluvial erosion and a probe of early Martian climate conditions, and by lava-ice interactions as a mechanism for creating relatively long-lived aqueous environments on Amazonian Mars. In Chapter 1, we used a snowpack energy balance model to investigate rates of snowmelt under a variety of warming scenarios for early Mars. Other workers (Irwin et al., 2005; Hoke et al., 2011) have presented estimates for the formative runoff rates in several valley networks based on the morphometry of preserved interior channels in these valleys. We showed that modeled snowmelt runoff rates consistent with these calculated runoff rates occur when greenhouse warming from an idealized gas is strong enough that annual average temperatures in the valley network catchments are approximately 270° – ix 280° K. Stronger warming leads to excessively rapid runoff rates and continuous fluvial activity throughout the year, both of which are inconsistent with geomorphologic evidence. Our results suggest that snowmelt as a mechanism to carve the valley networks may require much less greenhouse warming than rainfall. This is an important consideration given that a physical gas capable of warming early Mars to the required extent has not yet been identified. While most late Noachian – early Hesperian fluvial features are located in the equatorial highlands, in Chapter 2, we investigated the implications of a thicker, and possibly warmer, atmosphere for the climate at the south pole of Mars. The extensive and asymmetric Dorsa Argentea Formation has been interpreted as the remains of a large south circumpolar ice sheet (Head and Pratt, 2001). A large ice sheet in the south polar highlands is an expected consequence of a thicker atmosphere on early Mars because, unlike the tenuous modern atmosphere, such an atmosphere would be able communicate its heat to the surface, leading to an altitude control on surface temperature (Forget et al., 2013; Wordsworth et al., 2013). Using a global climate model, we found that the asymmetry of the ice sheet is likely to reflect the influence of the surface temperature dependence on altitude in the context of regional topography near the south pole. Using a glacial flow model, we found that greenhouse gases other than CO2 are required for the ice sheet to reach the observed extent of the DAF, and to induce basal melting in the regions where eskers, breached crater lakes, and fluvial channels are currently observed. Combined with crater exposure ages for the eskers (Kress and Head, 2015), this suggests that basal melting occurred in the DAF at the same time as widespread fluvial activity was occurring in the equatorial highlands. x In Scanlon et al. (2014), we documented evidence for several classes of glaciovolcanic landforms across the equatorial Arsia Mons fan-shaped deposit (FSD), which is interpreted as the remnants of a predominantly cold-based Late Amazonian glacier (Head and Marchant, 2003). In Chapter 3, we investigated the effects of these lava-ice interactions on the style of glaciation in the FSD, and presented a brief review of microbial habitats in terrestrial glaciovolcanic settings. We described several fields of ribbed moraine and a region of thrust-block moraine, both of which record transitions between wet-based and cold-based regions of a polythermal ice sheet. We synthesized the distribution of these features and the glaciovolcanic features documented in Scanlon et al. (2014) into a self-consistent description of a cold-based ice sheet where basal melting and sliding were enabled locally by lava-ice heat transfer during recent active volcanism. The local environments created by glaciovolcanic interactions, including subglacial lakes and wet basal sediments, may have been some of the largest and longest-lived aqueous environments on the surface of Mars in the late Amazonian. In Chapter 4, we documented additional previously undescribed glacial landforms in the FSD, and presented geomorphologic evidence that some ice from the Arsia Mons FSD glacier may still be buried in a subset of these landforms. We argued that a series of pits and knobs surrounding the edge of a remnant glacier near the northern edge of the deposit is a type of kame-and-kettle topography, with ice cores remaining in the knobs. We drew analogy between dunes in the FSD with axial troughs along their crests and terrestrial niveo-aeolian landforms. We documented similar cracks along the crests of some recessional moraines, as well as evidence for glaciovolcanic interactions between xi subaerial lava flows and stratigraphically lower recessional moraines, suggesting that these moraines also once contained ice cores and that a subset may still retain them. In Chapter 5, we review the contributions of this work to knowledge of the history of water ice on Mars. We propose several tests for the “icy highlands” model of late Noachian – early Hesperian Mars, and outline ways to build on our results. We also discuss the relevance of terrestrial glaciovolcanic environments as Mars analogues and suggest several avenues for future research aimed at increasing their applicability. References Forget, F., R. Wordsworth, E. Millour, J. B. Madeleine, L. Kerber, J. Leconte, E. Marcq, and R. M. Haberle (2013), 3D modelling of the early martian climate under a denser CO2 atmosphere: Temperatures and CO2 ice clouds, Icarus, 222(1), 81-99. Irwin, R. P., R. A. Craddock, and A. D. Howard (2005), Interior channels in Martian valley networks: Discharge and runoff production, Geology, 33(6), 489–492. Head, J.W., and D. R. Marchant (2003), Cold-based mountain glaciers on Mars: Western Arsia Mons, Geology, 31, 641-644. Head, J. W., and S. Pratt (2001), Extensive Hesperian-aged south polar ice sheet on Mars: Evidence for massive melting and retreat, and lateral flow and pending of meltwater, Journal of Geophysical Research: Planets, 106(E6), 12275-12299. Hoke, M. R. T., B. M. Hynek, and G. E. Tucker (2011), Formation timescales of large Martian valley networks, Earth and Planetary Science Letters, 312, 1–12. Kress, A. M., and J. W. Head (2015), Late Noachian and early Hesperian ridge systems in the south circumpolar Dorsa Argentea Formation, Mars: Evidence for two xii stages of melting of an extensive late Noachian ice sheet, Planetary and Space Science, 109, 1-20. Scanlon, K. E., J. W. Head III, L. Wilson, and D. R. Marchant (2014), Volcano-ice interactions in the Arsia Mons tropical mountain glacier deposits, Icarus, 237, 315-339. Wordsworth, R. D., Kerber, L., Pierrehumbert, R. T., Forget, F., & Head, J. W. (2015). Comparison of “warm and wet” and “cold and icy” scenarios for early Mars in a 3-D climate model. Journal of Geophysical Research: Planets, 120(6), 1201- 1219. xiii Table of Contents Title Page ............................................................................................................................. i Copyright ............................................................................................................................ ii Signature Page ................................................................................................................... iii Curriculum Vitae ............................................................................................................... iv Acknowledgements ............................................................................................................ vi Preface................................................................................................................................ ix CHAPTER 1. Modeled Snowmelt Rates in Early Martian Climate Scenarios........... 1 Abstract............................................................................................................................ 2 1.1 Introduction ............................................................................................................... 3 1.2 Snowmelt Model........................................................................................................ 6 1.3 GCM Simulations .................................................................................................... 10 1.4 Results ..................................................................................................................... 11 1.5 Discussion................................................................................................................ 16 1.6 Conclusions ............................................................................................................. 17 Acknowledgements ....................................................................................................... 19 References ..................................................................................................................... 19 Tables............................................................................................................................. 25 Figures ........................................................................................................................... 29 CHAPTER 2. The Dorsa Argentea Formation and the Noachian-Hesperian Climate Transition ........................................................................................................................ 57 Abstract.......................................................................................................................... 58 2.1. Introduction ............................................................................................................ 59 2.2. Estimates of subglacial melt fluxes and melt longevity from esker morphometry 67 2.3. Modeling the development and ablation of the DAF in an Early Mars GCM ....... 69 2.3.1 Climate conditions under which a large, asymmetric ice sheet can form......... 69 2.3.2 Ablation patterns and rates for the DAF ice sheet ............................................ 73 2.4 Modeling the DAF ice sheet with a Mars-adapted glacial flow model ................... 76 2.5 Discussion................................................................................................................ 82 2.6 Conclusions ............................................................................................................. 91 Acknowledgements ....................................................................................................... 92 References ..................................................................................................................... 92 Tables........................................................................................................................... 105 Figures ......................................................................................................................... 107 CHAPTER 3. Volcanism-Induced, Local Wet-Based Glacial Conditions Recorded in the Late Amazonian Arsia Mons Tropical Mountain Glacier Deposits .................. 163 Abstract........................................................................................................................ 164 xiv 3.1. Introduction .......................................................................................................... 164 3.2. Data....................................................................................................................... 167 3.3. Glacial landforms ................................................................................................. 168 3.3.1 Outflow channels, streamlined knobs, and drop moraines ............................. 168 3.3.2 Ribbed moraines.............................................................................................. 169 3.3.3 Thrust-block moraine ...................................................................................... 171 3.4 Discussion.............................................................................................................. 173 3.5 Potential microbial habitats in the Arsia Mons FSD ............................................. 174 3.5.1 Englacial and basal lakes ............................................................................... 175 3.5.2 Wet-based glacier ice ...................................................................................... 176 3.5.3 Glacial ice and debris ..................................................................................... 177 3.5.4 Hyaloclastite and palagonite .......................................................................... 178 3.5.5 Biosignature preservation potential ................................................................ 179 3.6 Conclusions ........................................................................................................... 180 Acknowledgements ..................................................................................................... 181 References ................................................................................................................... 181 Figures ......................................................................................................................... 194 CHAPTER 4. Remnant Buried Ice in the Equatorial Regions of Mars: Morphological Indicators Associated with the Arsia Mons Tropical Mountain Glacier Deposits ............................................................................................................ 209 Abstract........................................................................................................................ 210 4.1. Introduction .......................................................................................................... 211 4.2. Data and Methods ................................................................................................. 213 4.3. Landforms Interpreted to be Indicative of Remnant Ice....................................... 214 4.3.1 Pit-and-knob terrain........................................................................................ 214 4.3.2 Drop moraines with linear troughs ................................................................. 217 4.4 Postglacial Ice-Related Landforms ........................................................................ 217 4.5 Discussion.............................................................................................................. 220 4.6 Conclusions ........................................................................................................... 223 Acknowledgements ..................................................................................................... 224 References ................................................................................................................... 224 Figures ......................................................................................................................... 236 CHAPTER 5. Synthesis ................................................................................................ 254 Key geological predictions .......................................................................................... 258 Controls on age and development of valley networks .............................................. 258 Hellas and Argyre ice sheets .................................................................................... 260 Hellas and Argyre oceans ........................................................................................ 261 Future work.................................................................................................................. 262 References ................................................................................................................... 267 xv List of Tables CHAPTER 1. Modeled Snowmelt Rates in Early Martian Climate Scenarios........... 1 Table 1.1 Values adopted for snowmelt model parameters .......................................... 25 Table 1.2 Valley network names, locations, and calculated formative runoff ............. 26 Table 1.3 Annual average temperature at studied valley networks .............................. 27 Table 1.4 Annual cumulative rainfall at studied valley networks ................................ 28 CHAPTER 2. The Dorsa Argentea Formation and the Noachian-Hesperian Climate Transition ........................................................................................................................ 57 Table 2.1 Morphometry and estimated formative discharges for the Dorsa Argentea and Parva Planum esker populations ........................................................................... 105 Table 2.2 Time to construct Dorsa Argentea Formation eskers for varying estimates of ice debris content and formative discharge ................................................................. 106 xvi List of Figures CHAPTER 1. Modeled Snowmelt Rates in Early Martian Climate Scenarios........... 1 Figure 1.1. Example of an interior channel in Licus Vallis .......................................... 29 Figure 1.2. Sensitivity of results at Paraná Valles to albedo, snow density, snowpack surface roughness length, and snowpack depth ............................................................. 30 Figure 1.3. Map of annual average surface temperature across Mars at varying grey gas absorptivity .................................................................................................................... 32 Figure 1.4. Rolling 24-hour summed snowmelt rates (in m day-1) for Evros Vallis. .. 34 Figure 1.5. Rolling 24-hour summed snowmelt rates (in m day-1) for Licus Vallis.. . 37 Figure 1.6. Rolling 24-hour summed snowmelt rates (in m day-1) for Paraná Valles . 40 Figure 1.7. Rolling 24-hour summed snowmelt rates (in m day-1) for an unnamed valley network at 0°N, 23°E. ......................................................................................... 43 Figure 1.8. Rolling 24-hour summed snowmelt rates (in m day-1) for an unnamed valley network at 6.6°S, 135°E...................................................................................... 46 Figure 1.9. Comparison of annual average GCM surface temperatures at varying grey gas absorptivity, showing locations of study sites......................................................... 49 Figure 1.10. Snowmelt rates and, where applicable, rainfall rates at each study site... 51 Figure 1.11. Map showing annual cumulative rainfall (in m year-1) at all study sites with 1000 mb CO2 ......................................................................................................... 54 Figure 1.12. Map showing surface water ice distribution (in kg m-2) after ~400 model years of ice evolution with a 1000 mb pure CO2 atmosphere ....................................... 55 CHAPTER 2. The Dorsa Argentea Formation and the Noachian-Hesperian Climate Transition ........................................................................................................................ 57 Figure 2.1. Geologic units and features of the Dorsa Argentea Formation ................ 107 Figure 2.2. Schematic representation of possible trajectories for the development of DAF landforms ............................................................................................................ 108 Figure 2.3. Annual average surface temperature (°K) in the southern hemisphere with varying global CO2 surface pressure and spin-axis obliquity of 25°........................... 109 Figure 2.4. Annual average surface temperature (°K) in the southern hemisphere with varying global CO2 surface pressure and spin-axis obliquity of 41.8°........................ 111 Figure 2.5. Annual average surface temperature (°K) in the southern hemisphere with varying global topography and spin-axis obliquity of 25°. ......................................... 113 Figure 2.6. Annual average surface temperature as a function of altitude and surface pressure in GCM simulations with 25° and 41.8° spin-axis obliquity. ....................... 115 Figure 2.7. GCM surface temperature (°K) contours superimposed on geologic map showing the Dorsa Argentea eskers............................................................................. 116 Figure 2.8. Annual maximum surface temperatures (°K) in a GCM simulation with a 1500 mb CO2 atmosphere and 41.8° spin-axis obliquity. ........................................... 117 Figure 2.9. Ice accumulation and ablation after ten model years in the region of the DAF at 25° obliquity and varying CO2 surface pressure. ........................................... 118 Figure 2.10. Ice accumulation and ablation after ten model years in the region of the DAF at 41.8° obliquity and varying CO2 surface pressure. ........................................ 120 xvii Figure 2.11. Ice accumulation and ablation after ten model years in the region of the DAF with increased solar flux and with varying obliquity and CO2 surface pressure.122 Figure 2.12. Accumulation and ablation of a flat ice sheet with the spatial extent of the DAF after ten model years at varying obliquity and CO2 surface pressure. ............... 123 Figure 2.13. Accumulation and ablation of a flat ice sheet with the spatial extent of the DAF after ten model years at low atmospheric pressure. ............................................ 125 Figure 2.14. Accumulation and ablation of a flat ice sheet with the spatial extent of the DAF after ten model years in a 50 mb CO2 atmosphere with ice albedo set equal to ground albedo. ............................................................................................................. 126 Figure 2.15. Location of the ~20 m ice contour after ~100 – ~400 years of ice ablation under increased solar flux. ........................................................................................... 127 Figure 2.16. Equilibrium surface topography (in m) of the modeled DAF ice sheet in simulations under varying CO2 surface pressure. ........................................................ 128 Figure 2.17. Equilibrium surface topography (in m) of the modeled DAF ice sheet in simulations with varying global ice inventory. ........................................................... 129 Figure 2.18. Equilibrium surface topography (in m) of the modeled DAF ice sheet in simulations under varying geothermal heat flux. ........................................................ 131 Figure 2.19. Flow velocity (in log10 mm yr-1) of the modeled DAF ice sheet in simulations under varying CO2 surface pressure. ........................................................ 132 Figure 2.20. Basal melting rate (in mm yr-1) of the modeled DAF ice sheet in simulations under varying CO2 surface pressure. ........................................................ 133 Figure 2.21. Basal melting rate (in mm yr-1) of the modeled DAF ice sheet in simulations with varying global ice inventory. ........................................................... 134 Figure 2.22. Basal melting rate (in mm yr-1) of the modeled DAF ice sheet in simulations under varying geothermal heat flux. ........................................................ 136 Figure 2.23. Basal temperature (in °K) of the modeled DAF ice sheet in simulations under varying CO2 surface pressure. ........................................................................... 137 Figure 2.24. Basal temperature (in °K) of the modeled DAF ice sheet in simulations with varying global ice inventory. ............................................................................... 138 Figure 2.25. Difference (°K) between annual average surface temperatures in the GCM simulations with various “grey gas” absorptivities and 1000 mb CO2........................ 140 Figure 2.26. Equilibrium surface topography (in m) of the modeled DAF ice sheet in simulations with varying grey gas absorptivity. .......................................................... 143 Figure 2.27. Equilibrium surface topography (in m) of the DAF ice sheet in simulations climate data from a GCM run with a 1000 mb atmosphere and grey gas κ = 1.0 × 10-4, with varying global ice inventory. ............................................................. 148 Figure 2.28. Basal melting rate (in mm yr-1) of the modeled DAF ice sheet in simulations with varying grey gas absorptivity. .......................................................... 151 Figure 2.29. Basal melting rate (in mm yr-1) of the DAF ice sheet in simulations climate data from a GCM run with a 1000 mb atmosphere and grey gas κ = 1.0 × 10-4, with varying geothermal heat flux. .............................................................................. 156 Figure 2.30. GCM simulated surface water ice, with the areal centroid of the DAF indicated by a black square. ......................................................................................... 159 Figure 2.31. Hypothesized mechanisms for generating the meltwater that flowed beneath the DAF ice sheet and allowed eskers to form. .............................................. 160 xviii Figure 2.32. Possible sites of explosive glaciovolcanism associated with the Dorsa Argentea eskers............................................................................................................ 162 CHAPTER 3. Volcanism-Induced, Local Wet-Based Glacial Conditions Recorded in the Late Amazonian Arsia Mons Tropical Mountain Glacier Deposits .................. 163 Figure 3.1. Geomorphological unit map of the Arsia Mons fan-shaped deposit (FSD), after Zimbelman and Edgett (1992) and Scott and Zimbelman (1995). ...................... 194 Figure 3.2. Image mosaic and sketch map of the Northwest Plateau. ........................ 195 Figure 3.3. Close view of drop moraines downslope of the Northwest Plateau. ........ 197 Figure 3.4. Map view schematic showing formation of outward-bowed moraines in response to locally enhanced glacial flow. .................................................................. 198 Figure 3.5. Regional context map showing contacts between the Lobate, Ridged, and Knobby facies. ............................................................................................................. 199 Figure 3.6. Concentric, sharp, evenly spaced ridges, interpreted as ribbed moraines, at the contact between the Lobate Facies and the Knobby Facies. ................................. 200 Figure 3.7. Schematic of ribbed moraine formation where a glacier transitions from wet-based to cold-based conditions. ............................................................................ 201 Figure 3.8. Ridges interpreted as ribbed moraines on the Northwest Plateau. ........... 202 Figure 3.9. The L-shaped ridge and surrounding landforms. ..................................... 203 Figure 3.10. Terminal moraine near the L-shaped ridge, showing ribbed surface appearance. .................................................................................................................. 204 Figure 3.11. Topographic profiles across terminal moraine. ...................................... 205 Figure 3.12. Schematic of formation for proposed thrust-block moraine. ................. 207 Figure 3.13. Summary schematic showing the relationship of inferred glaciovolcanic and wet-based / polythermal glacial landforms in the Arsia Mons FSD. .................... 208 CHAPTER 4. Remnant Buried Ice in the Equatorial Regions of Mars: Morphological Indicators Associated with the Arsia Mons Tropical Mountain Glacier Deposits ............................................................................................................ 209 Figure 4.1. Geomorphological unit map of the Arsia Mons fan-shaped deposit (FSD), after Zimbelman and Edgett (1992) and Scott and Zimbelman (1995). ...................... 236 Figure 4.2. Pit and knob terrain in the Arsia Mons fan-shaped deposit. .................... 237 Figure 4.3. Image and sketch map showing individual pits and knobs. ..................... 238 Figure 4.4. Examples of knobs, moated knobs, pits with remnant knobs, and pits. ... 239 Figure 4.5. Pits and knobs fall along lines concentric to the boundaries of the Smooth Facies. .......................................................................................................................... 241 Figure 4.6. Schematic comparison of kettle hole formation process on Earth and a “backwasting” model of pit and knob terrain formation in the Arsia Mons fan-shaped deposit.......................................................................................................................... 242 Figure 4.7. Schematic of a “controlled moraine” model of pit-and-knob terrain formation in the Arsia Mons fan-shaped deposit......................................................... 243 Figure 4.8. Candidate ice-cored moraines. ................................................................. 245 Figure 4.9. Typical dunes with linear troughs, found throughout the western extent of the Arsia Mons fan-shaped deposit. ............................................................................ 247 xix Figure 4.10. Field of dunes with linear troughs, showing typical distribution and orientation. ................................................................................................................... 248 Figure 4.11. Schematic dunes with linear troughs forming from dust-ice dunes. ...... 249 Figure 4.12. Schematic dunes with linear troughs forming from dust-covered snow dunes. ........................................................................................................................... 251 Figure 4.13. Dust- and sand-covered ripples in the McMurdo Dry Valleys. ............. 252 Figure 4.14. Topography of a region of pit and knob terrain. .................................... 253 xx Chapter 1 Modeled Snowmelt Rates in Early Martian Climate Scenarios Kathleen E. Scanlon and James W. Head Department of Earth, Environmental and Planetary Sciences, Brown University 324 Brook Street, Box 1846, Providence, RI, 02912, USA 1 Abstract. The paleodischarge of water through a valley can be calculated from the morphometry of small interior channels where they are preserved. To determine likely runoff rates from snowmelt under a variety of scenarios for early Mars, and assess their agreement with runoff rates calculated by other workers from observed channel dimensions, we conducted a series of global climate model (GCM) simulations of early Mars, inducing climate warming with a grey gas (i.e. an artificial, wavelength- independent absorption coefficient) of varying strength. We used the output from these simulations as input to an energy balance snowmelt model, and compared the modeled runoff rates at five valley networks to those that have been calculated from interior channel morphometry by other workers. We found that runoff consistent with observations can occur from snowmelt alone when greenhouse warming is strong enough that annual average temperatures in the highlands are approximately 270° – 280° K. Excessive warming leads to both large runoff rates and continuous melting throughout the year, both of which are inconsistent with geomorphologic evidence. Sufficient runoff rates can be achieved at sites where maximum rainfall rates are well below those inferred from interior channel morphometry, which suggests that snowmelt as a mechanism to carve the valley networks may require much less greenhouse warming than rainfall. Substantial warming above the “Late Noachian Icy Highlands” baseline is still required, however, and the physical mechanism for this warming remains unknown. 2 1. Introduction The dendritic valley networks dissecting the equatorial highlands of Mars (e.g. Fassett and Head, 2008; Irwin, 2008; Hynek et al., 2010) are generally thought to record a warmer era in martian climate history, when liquid water could flow across the surface long enough and in sufficient quantity to erode valleys as large as hundreds of kilometers long (Hynek et al., 2010) and averaging ~2 km wide and 100 m deep (Williams and Phillips, 2001). The dense, well-developed networks visible in high-resolution remote sensing images (e.g. Hynek and Phillips, 2003) suggest that surface runoff, rather than groundwater sapping, is responsible for most valley networks. Whether this runoff originated from rainfall (e.g. Craddock and Howard, 2002; Hoke and Hynek, 2009; Grant et al., 2011) or from intermittent melting of snow and ice accumulated during colder periods (e.g. Wordsworth et al. 2013, 2015; Head and Marchant, 2014) remains a matter of debate. Craddock and Howard (2002) argued that an origin from rainfall is more likely, on the basis of the profiles of degraded Noachian craters most strongly resembling modeled profiles of craters degraded by rainfall and surface runoff. The lack of a plausible composition for the early martian atmosphere that would have provided enough greenhouse warming to compensate for the faint young sun and sustain rainfall (e.g., Haberle, 1998; Tian et al., 2010; Forget et al., 2013), however, has motivated a search for alternative explanations. Furthermore, recent modeling indicates that, unlike the likely regional-scale pattern of orographic enhancement (Scanlon et al., 2013) or the large-scale pattern of snow and ice preservation on a cold early Mars (Wordsworth et al., 2013; 2015), the pattern of rainfall in a hypothetical warm early martian climate may not have been consistent with the distribution of valley networks (Wordsworth et al., 2015). 3 Because the small interior channels on valley floors (Figure 1) record the widths of the fluvial flows that eroded the valleys, the discharge rate of water through a valley network can be calculated from their morphometry where they are preserved. Irwin et al. (2005) adapted terrestrial scaling relationships between channel discharge and channel width to be appropriate to Mars. They used the Mars-adapted relationships to estimate that the valley networks with preserved channels formed at discharges of ~300 – 3,000 m3 s-1, corresponding to runoff rates ranging from 0.1 cm day-1 for networks in Sinus Sabaeus to 5.5 cm day-1 for a network in Mare Tyrrhenum. Hoke et al. (2011) used valley network morphometry, physically based equations for flow depth and flow velocity, and reasonable estimates for martian parameters such as sediment density and grain size to calculate that the valley networks in their regions of study formed under discharges of 7,000 – 30,000 m3 s-1, corresponding to runoff rates ranging from 0.4 – 5 cm day-1. For terrestrial comparison, maximum rainfall rates in thunderstorms near Kennedy Space Center range between 5 mm hour-1 and ~18 cm hour-1 (Piepgrass et al., 1982), most of the 2.5 - 3 cm of rainfall each year in the Negev Desert occurs as a few brief storms with rainfall rates of at least 2 - 3 cm hour-1 (Amit et al., 2010), and most storms in the Mojave, Chihuahuan and Sonoran Deserts leave 0.5 cm of rain or less (Reynolds et al., 2004). Barnhart et al. (2009), using a martian landscape evolution model, estimated discharges for Loire, Samara, and Paraná Valles comparable to those Irwin et al. (2005) and Hoke et al. (2011) calculated from interior channel morphometry. These estimated runoff rates may provide one means of constraining the source of the water that carved the Noachian valley networks. The discharge of even the largest melt-fed streams in the present-day Antarctic Dry Valleys is orders of magnitude smaller 4 than discharges in the martian valley networks (e.g. Head and Marchant, 2014; Levy, 2015), but springtime runoff rates from snowmelt in terrestrial subpolar climates can exceed the rates inferred for the valley networks (e.g. Dunne et al., 1976; Puurveen et al., 1997; Janowicz et al., 2004), to the extent that snowmelt in rapidly warming modeled martian climates may even generate too much runoff to be consistent with some valley networks. To determine the likely magnitude of runoff from snowmelt under a variety of scenarios for early Mars, we conducted a series of global climate model (GCM) simulations of early Mars in a cold and icy scenario, then induced climate warming by adding an artificial, wavelength-independent absorption coefficient κ (i.e. “grey gas”). We used the output from these GCM simulations at the approximate locations of a subset of the valley networks studied by Irwin et al. (2005) and Hoke et al. (2011) as input to an energy balance snowmelt model, and compared the resulting runoff to required runoff rates calculated from valley network morphometry. We investigated four key questions: (1) Can snowmelt alone generate runoff of comparable magnitude to the runoff estimated from channel morphometry in several valley networks? (2) Can snowmelt rates on the order of mm day-1 to cm day-1 be reached in regions where annual average surface temperatures are below freezing and peak rainfall rates are lower than 1 mm day-1? (3) Is there a single climate scenario where modeled snowmelt runoff best matches calculated runoff for all valley networks studied? (4) Are modeled snowmelt rates larger in the valley networks with larger calculated runoff rates relative to the others? 5 2. Snowmelt model The Utah Energy Balance (UEB) snowmelt model (Mahat and Tarboton, 2014; Tarboton and Luce, 1996) characterizes the depth, energy content, and surface age of a snowpack by solving the energy and mass balance equations !" = !!" + !!" + !! + !! − !!" + !! + !! − !! !" and !" !" = !! + !! + !! − !, where Qsn is net shortwave radiation, Qli is incoming longwave radiation, Qp is heat advection into the snowpack by precipitation, Qg is ground heat flux, Qle is outgoing longwave radiation, Qh is sensible heat flux, Qe is latent heat flux, Qm is heat advection out of the snowpack by meltwater, Pr is rainfall rate, Ps is snowfall rate, Mr is meltwater outflow rate, and E is the sublimation rate of snow (Tarboton and Luce, 1996). The model takes as inputs temperature, precipitation, wind speed, relative humidity, and shortwave and longwave radiation at the surface, as well as annual average surface pressure. The purpose of this study is to evaluate the rates at which snow, having collected and compacted into firn and ice over cold periods in between warming events (in the “Late Noachian Icy Highlands” model, Wordsworth et al. 2013), could melt upon the introduction of warming. The values we used for the snowpack thermal and physical parameters (Table 1) reflect this aim, and are based upon the assumption that cold periods were at least a few thousand years long. Based on estimates for the late Noachian ice inventory (Carr and Head, 2015), snow accumulation rates in the highlands during the late Noachian (Wordsworth et al., 6 2013), and the equilibrium distribution of ice in the late Noachian highlands (Fastook and Head, 2014), Cassanelli and Head (2015) note that late Noachian ice sheets might have been deposited in a few tens of kyr and persisted without additional deposition until melting (and hence hydrologic recycling) could occur. They modeled densification of ice sheets during these quiescent periods and found that the depth of the firn-to-ice transition shallows from ~110 m to ~20 m over the course of 1 Myr. Following these results, we adopted a snowpack thickness of 20 m for our final set of simulations, and ran the model assuming a glacial substrate underneath snowpack; melt model parameters such as ground heat capacity and density were set to values for water ice. We set the thermal and physical properties of the snowpack to moderate values from Cassanelli and Head (2015). Initial snowpack heat content was set to give an average initial snowpack temperature of 250°K, consistent with the initial conditions of the GCM simulation (section 3). In order to isolate runoff from snowmelt without the effects of advected heat from rainfall, precipitation was set to fall entirely as snow. Other snowmelt model parameters were set to reasonable estimates based on values from Earth and modern Mars. For example, visual properties of the snowpack (e.g. albedo of fresh snow, emissivity) were taken from Warren (1982), and snowpack surface aerodynamic roughness was set to 0.002 m, appropriate for the present-day martian polar caps (Heavens et al., 2008). While most of the snowmelt model parameters are reasonably well-constrained by the assumption of a water ice substrate, the visual albedo of the snowpack, aerodynamic roughness length of the snow surface, and depth of the firn-to-ice transition could all vary across a wide range due to the uncertainty (and likely variability) in snow debris content 7 and time between warm intervals. Snowpack density is treated as a constant by the UEB model, but in the scenario considered here, where snow accumulates for long periods of time in between melting episodes, snowpack density is likely to vary with depth (e.g. Cassanelli and Head, 2015). Finally, to our knowledge there is no non-empirical expression for snowpack liquid holding capacity, and values for Mars are therefore difficult to constrain. We qualitatively assessed the sensitivity of the model to these five parameters, illustrated using the 25° obliquity, 1000 mb CO2, κ = 2 × 10-4 m2 kg-1 simulation at Paraná Valles. Albedo. On Earth, snowpack albedo is one of the most sensitive model parameters (Tarboton and Luce, 1996). Reducing albedo from 0.9 to 0.6 in the early Mars model, however, does not substantially affect the timing of snowmelt and only changes its magnitude by a few percent (Figure 2a). This is likely to be an effect of the lower insolation on early Mars relative to modern Earth, due to both the faint young sun and the greater distance between Mars and the Sun. Snow density. The assumed density of the modeled snowpack also has only a minor effect on melt generation (Figure 2b); changing density across values representing the range from freshly fallen snow to dense firn (Cassanelli and Head, 2015), 100 kg m-3 to 800 kg m-3, only slightly changes the magnitude and timing of a few late-season peaks. Roughness length. Results are not strongly sensitive to the aerodynamic roughness length z0 between 0.0001 and 0.001 m, but snowmelt magnitude is substantially greater at z0 = 0.01 m (Figure 2c). On Earth, roughness lengths over glacier snow are typically of order 0.0001 – 0.001 m (see summary in Brock et al., 2009); only 8 very rough glacier ice has z0 of order 0.01 m, and this paper investigates the melt of snow and firn on a glacier ice substrate rather than the melting of the ice itself. Snowpack depth. Melt generation in the model is sensitive to snowpack depth only at much smaller depths than we consider in this study. The difference between snowmelt in the same simulation with a 20 m or 100 m thick snowpack is of the order of hundredths of a percent for most of the year, but increases as melt continues and the size of the snowpack diminishes (Figure 2d). Liquid holding capacity. Varying the liquid holding capacity between 5% and 15% produced no change in our results. This is likely to be because melting is continuous throughout the 25° obliquity, 1000 mb CO2, κ = 2 × 10-4 simulation at Paraná Valles, leaving the snowpack continuously saturated regardless of holding capacity. When the UEB model is used in terrestrial applications, field measurements provide these inputs, but since field data are obviously not available for Noachian Mars, we use simulated climate data from the Laboratoire de Météorologie Dynamique (LMD) early Mars GCM (Forget et al., 2013; Wordsworth et al., 2013) as input instead. Specifically, surface long-wave and short-wave radiative flux, surface air temperature, and near-surface wind speed and relative humidity were collected from the GCM on an hourly basis. Output from the snowmelt model includes snow energy content, snow water equivalence, sublimation rate, and melt outflow rate; here we focus on the magnitude and timing of melt outflow. The UEB model is physics-based, using empirical parameterizations only when a desired climate variable is not available from direct measurements at a terrestrial research site. Since we use GCM-simulated data for all input climate variables, no terrestrial 9 empirical parameterizations were necessary, and only changes to constants (namely, gravitational acceleration, atmospheric heat capacity, and atmospheric specific gas constant) were necessary. We chose five valley networks from Irwin et al. (2005) and Hoke et al. (2011) to represent a range in spatial distribution and calculated runoff magnitude. The names, locations, and calculated runoff for the valley networks studied are given in Table 2. 3. GCM Simulations The LMD Generic GCM allows users to simulate the climate of any planet by varying the modeled planet’s size, gravity, orbital parameters, and atmospheric composition. Using parameters for Early Mars, and a model resolution of 64 grid points in longitude, 48 grid points in latitude, and 25 vertical levels, we initialized the GCM with an isothermal 600 or 1000 mb CO2 atmosphere at 250°K and 25° or 41.8° spin-axis obliquity. The initial surface distribution of water ice was set to ~50 m of ice at every grid point with surface elevation of at least 1000 m; this altitude was chosen to approximately delineate the southern highlands from the northern lowlands and the Hellas and Argyre basins. Since a physically plausible combination of greenhouse gases that could have warmed broad regions of the equatorial highlands has not yet been identified (e.g. Haberle, 1998; Forget et al., 2013; von Paris et al., 2013; Kerber et al., 2014) we added an idealized grey gas in our GCM simulations to simulate a sudden influx of greenhouse gases into the atmosphere, as if by volcanism. The first year of GCM data from each scenario was used as input to the snowmelt model. 10 We chose κ = 2.5 × 10-5, 5 × 10-5, 1 × 10-4, and 2 × 10-4 m2 kg−1 in order to study a range of warmer climates; at our warmest study site, for example, temperatures only exceeded freezing for a few tens of sols at κ = 2.5 × 10-5 m2 kg−1, but exceeded 273°K by ~10-20° for most of the year at κ = 2.5 × 10-4 m2 kg−1. The varied absorption coefficients here are not meant to represent specific gases or concentrations of gases that have been proposed as possible agents of warming on early Mars, since a grey gas has constant absorption properties across the infrared spectrum. The absorption coefficients of real gases vary by several orders of magnitude across the infrared spectrum, even over finite ranges such as the wavenumber ~1100 – ~1800 cm-1 “window” where CO2 absorbs weakly (and hence where the efficacy of a greenhouse gas for early Mars may largely be determined). As an empirical illustration of their relative effectiveness, however, Figure 3 shows the annual average surface temperature of Mars for each value of κ in the 25°, 1000 mb CO2 simulations. For comparison, von Paris et al. (2013) found that 500 mb of N2 in a ~400 mb – 1000 mb CO2 atmosphere could warm the surface by at least 12°K, and Kerber et al. (2014) found that adding 100 ppmv of SO2 to a 500 mb CO2 atmosphere warmed the martian highlands by 10-20°K. 4. Results Figures 4 – 8 show modeled snowmelt as a function of time for each valley network. For direct comparison with the melt rates calculated by Irwin et al. (2005) and Hoke et al. (2011), we present our results as running 24-hour melt totals (units m day-1). Modeled melt rates ranged from hundreds of micrometers through ~15 centimeters per day, with strong seasonal and diurnal peaks. After the initial warming, snowmelt 11 continued year-round (or until the snowpack was exhausted) in our warmest simulations (Patm = 1000 mb, κ = 2 × 10-4 m2 kg−1), where annual average temperature was above freezing for all points except Licus Vallis. In simulations with κ = 1 × 10-4 m2 kg−1, most melt occurred during two seasonal peaks 50 – 150 sols long. In simulations with κ = 5 × 10-5 m2 kg−1, when snowmelt occurred at all, it occurred only for intervals of a few sols at Evros Vallis and tens of sols at Paraná Valles. In simulations with κ = 2.5 × 10-5 m2 kg−1, snowmelt occurred only at Paraná Valles and only with Patm = 1000 mb. The first goal of our study was to determine whether snowmelt alone can generate runoff comparable to the formative runoff estimated from channel morphometry in the valley networks studied. When average atmospheric pressure was set to 600 mb, even warming by the grey gas with κ = 2 × 10-4 m2 kg−1 was insufficient to generate the required snowmelt rates, except at Paraná Valles and Evros Valles. In the warmest climates, however, with 1000 mb atmospheric pressure and grey gas κ = 2 × 10-4 m2 kg−1, snowmelt rates were faster than those inferred from observations at all five valley networks studied. In Paraná Valles and Evros Valles, excessive snowmelt was also generated in climates with 1000 mb CO2 and a grey gas with κ = 1 × 10-4 m2 kg−1. Melt rates with Patm = 1000 mb and κ = 5 × 10-5 m2 kg−1 were insufficient at Evros Vallis in the simulations with 25° spin-axis obliquity, but comparable to values calculated from channel morphometry at Evros Vallis when obliquity was 41.8°, and at Paraná Valles for both obliquity values. For Licus Vallis and the unnamed valley networks, insufficient snowmelt occurred with with κ = 1 × 10-4 m2 kg−1, and excessive snowmelt occurred with with κ = 2 × 10-4 m2 kg−1. Modeled snowmelt at Licus Vallis was less than calculated runoff for climates with κ < 2 × 10-4 m2 kg−1, and exceeded calculated runoff for κ = 2 × 12 10-4 m2 kg−1. Clearly, modeled runoff rates from snowmelt alone can meet—and exceed—the runoff rates calculated for the valley networks in warmer hypothetical martian climates. Our second goal was to determine whether snowmelt of the required magnitude could occur in regional climates with annual average temperatures below freezing and where rainfall alone is not sufficient to explain the valley networks. Figure 9 shows the global distribution of annual average temperature for each value of κ in the 25°, 600 mb GCM simulation, with the locations of the studied valley networks marked. Tables 3 and 4 list the annual average temperature and annual cumulative rainfall at each of our sites for all four GCM simulations; simulations where modeled snowmelt meets or exceeds calculated runoff are highlighted in dark grey. At Evros Vallis in the P = 600 mb, o = 41.8°, κ = 2 × 10-4 m2 kg−1 simulation; Licus Vallis in the P = 1000 mb, o = 25°, κ = 2 × 10-4 m2 kg−1 simulation; Paraná Valles in the P = 1000 mb, o = 25°, κ = 5 × 10-5 m2 kg−1 simulation; and Evros Vallis in the P = 1000 mb, o = 41.8°, κ = 2 × 10-4 m2 kg−1 simulation, sufficient runoff was generated with annual average temperatures a few degrees below freezing. Elsewhere, annual average temperatures exceeded freezing at all sites where modeled snowmelt met or exceeded calculated runoff. Only in our second warmest set of simulations (κ = 1 × 10-4 m2 kg−1, Patm = 1000 mb), and only at Evros Vallis, is rainfall alone sufficient to explain the runoff rates calculated from channel morphometry (Figure 10a). In the κ = 1 × 10-4 m2 kg−1 simulation for Unnamed Valley Network #1 (Figure 10d) and the κ = 2 × 10-4 m2 kg−1 simulation for Licus Vallis (Figure 10b), rainfall is a significant component of total runoff, but the rainfall is less than the runoff required by channel morphometry in the 13 former case, and snowmelt alone is more than required by channel morphometry in the latter case. At Paraná Valles and the unnamed valley networks, the annual cumulative rainfall rate (Table 4) is so low in all simulations that it would be insufficient to generate the required runoff even if the total annual rainfall occurred in one day. Unless the martian climate at the time of valley network formation was even warmer and wetter than in our κ = 2 × 10-4 m2 kg−1 GCM simulation, rainfall alone is therefore unlikely to account for the runoff in the valley networks. Our final goal was to assess whether the variations in snowmelt amounts between networks correlated with the variations in calculated runoff amounts, and whether a single simulation was most nearly suitable for all networks studied. In all of our simulations, regardless of atmospheric pressure, added grey gas absorption, or spin-axis obliquity, Licus Vallis consistently experienced the smallest peak snowmelt rates in our simulations, followed by Evros Vallis, consistent with the calculated runoff rates. Modeled snowmelt rates for Paraná Valles, however, were similar to those for the unnamed valley networks, which are calculated to have formed under runoff rates ~5x as rapid as that at Paraná. Potential reasons for the modeled runoff rates at the unnamed valley networks failing to exceed those at Evros, Licus and Paraná to the extent indicated by observations include (1) variations in soil infiltration capacity across the study sites; (2) variations in the amount of snow and ice built up between warm episodes at each study site, and hence the maximum total melt that can occur at each site; and (3) variations in rainfall, which could supplement runoff from snowmelt. Attempting to determine the soil infiltration capacity at the sites during the late Noachian is beyond the scope of this study, but Figure 14 11 shows the annual cumulative rainfall at all study sites with 1000 mb CO2, and grey gas κ = 5 × 10-5 m2 kg−1 (Figure 11a), κ = 1 × 10-4 m2 kg−1 (Figure 11b), and κ = 2 × 10-4 m2 kg−1 (Figure 11c), and Figure 12 shows the ice distribution in the model after ~400 years of ice evolution in the cold, CO2-only climate state at spin-axis obliquity 25° (Figure 12a) or 41.8° (Figure 12b). With moderate warming (κ = 5 × 10-5 or 1 × 10-4 m2 kg−1) and at 25° obliquity, more rainfall occurs at Unnamed Valley Network #1 than at Evros, Licus, or Paraná. This is not robust with respect to obliquity or warming magnitude even for Unnamed Valley Network #1, however. Furthermore, variations in rainfall cannot explain the lower-than- expected modeled runoff at Unnamed Valley Network #2 relative to Evros, Licus, and Paraná, as those four sites consistently experience similar annual rainfall totals to each other at both obliquities and all four values for κ. Variation in cold-interval ice accumulation between the five sites (Figure 12) is also unlikely to explain the inconsistencies in the calculated and modeled runoff trends, as the variability in ice accumulation between sites is minor relative to variability across the surface of the planet. There is no guarantee that all preserved interior channels date to the same episode of climate warming, however. Our favored explanation for the modeled runoff at Paraná exceeding that at the unnamed valley networks (despite calculated runoff having the opposite relationship) is that, due to its lower altitude and hence warmer climate, Paraná Valles remained active longer than the unnamed valley networks. Irwin et al. (2005) noted that the interior channels in their study appeared to record the abrupt end of fluvial activity in their respective valleys, which is consistent with this hypothesis. 15 5. Discussion In our warmest simulations, snowmelt is continuous throughout the year at all locations studied. Not only do melt rates in these simulations exceed the rates calculated for their respective valley networks, but the duration of melt events is also inconsistent with geomorphological evidence. Specifically, Barnhart et al. (2009) found that the relatively few crater breaches in the fluvial network are inconsistent with frequent fluvial activity in Paraná basin throughout the martian year, and that the morphology of the Paraná drainage network is consistent with high evaporation rates (and long periods of evaporation and/or groundwater infiltration between runoff events) and the presence of a moderately indurated surface crust (developed during periods of low-intensity runoff, and eroded by episodic floods). Our results suggest that if the valley networks formed in a climate as warm as our κ = 2 × 10-4 m2 kg-1 simulations, snowmelt rates would not only have exceeded the runoff rates inferred from interior channel morphometry, but would have done so almost continuously throughout the year. Therefore, if Mars was episodically warmed, either the magnitude of warming must have been less than in these simulations, snow accumulation between periods of rainfall must have been minimal (such that melting was brief even if warm temperatures were prolonged), or warm periods must have been extremely brief. An important limitation of this study is the relationship between snowmelt generation and actual runoff production. The infiltration capacity of soil in the valley network catchments at the time of their formation is unknown, and we have assumed for the purposes of this study that all snowmelt runs off. This is likely to be a fair approximation in a climate, such as a cold and icy early Mars warmed only intermittently 16 by volcanism or impacts, where temperatures rarely exceed freezing. In the Antarctic Dry Valleys, for example, melt infiltrates soil only to very shallow depths (tens of cm) due to the presence of an impermeable, ice-cemented permafrost layer below that depth (e.g. Conovitz et al., 2006; Bockheim et al., 2007; Levy et al., 2011; Head and Marchant, 2014). Due to the unconstrained state of dust in the Noachian atmosphere, we have not considered the effects of dust on snowpack thermal properties in this work. The presence of small amounts of dust can significantly decrease snowpack albedo, increasing the capacity of the snow to absorb solar radiation, but can also form a protective lag, inhibiting melt (Williams et al., 2008). Volcanic tephra, similarly, can either armor underlying snow and ice against sublimation or enhance melting by lowering surface albedo (Wilson and Head, 2009). Albedo variations may be less important for an intermittently warmed “icy highlands” early Mars (e.g. Figure 2a) due to the combined effects of a faint young sun, the strong greenhouse warming required to bring surface temperatures near the melting point, and rapid heat transport in a thicker atmosphere. Adapting the snowmelt model to explicitly model lag formation and the effects of dust on snowpack thermal properties for future work will nonetheless be illustrative. 6. Conclusions We find that snowmelt on early Mars can occur at rates comparable to the formative runoff rates estimated for several martian valley networks. This only occurs in climates warmed by a greenhouse gas strong enough that annual average temperatures are at least 270°-280°K in the regions incised by valley networks (Figure 3), however; 17 significant warming above the “icy highlands” baseline is therefore required, and the mechanism for this warming is still unknown. Snowmelt rates reach sufficient magnitude within tens of days, such that time-limited warming mechanisms such as crater impacts (e.g. Segura et al., 2002) or sulfur dioxide (e.g. Halevy and Head, 2014) would only need to remain in the warming phase for short periods of time if the magnitude and recurrence of warming was sufficient. Simulated rainfall rates are lower than snowmelt rates by at least one order of magnitude except for two regions in the κ = 1 × 10-4 m2 kg−1 simulation (Unnamed Valley Network #1 and Evros Vallis), indicating that snowmelt is likely to have been an important factor in valley network development even if part of the fluvial activity was due to rainfall. Our results are inconsistent with the valley networks forming in a climate as warm as our warmest simulation. This is due to (1) snowmelt rates in our warmest simulation exceeding the formative rates calculated for the valley networks and (2) snowmelt occuring continuously throughout the year in our warmest simulations, which is inconsistent with geomorphologic evidence for intermittent fluvial activity (e.g. Barnhart et al., 2009). Trends in snowmelt rates between valley networks correspond with trends in calculated formative runoff rates, with some exceptions; these exceptions may reflect the influence of varying snowpack buildup between warm intervals, rainfall distribution in warmer climates, varying substrate properties affecting infiltration capacity, or age variations between the channels. Future work should consider the potential effects of snowpack dust content and catchment infiltration capacity on these findings. 18 Acknowledgements We thank the NASA Graduate Student Researchers Program (Grant NNX12AI39H) for financial support to KES, and Robin Wordsworth for substantial and ongoing input to this chapter. References Amit, R., Y. Enzel, T. Grodek, O. Crouvi, N. Porat, and A. Ayalon (2010), The role of rare rainstorms in the formation of calcic soil horizons on alluvial surfaces in extreme deserts, Quaternary Research, 74, 177–187. Barnhart, C. J., A. D. Howard, and J. M. Moore (2009), Long-term precipitation and late- stage valley network formation: Landform simulations of Parana Basin, Mars, Journal of Geophysical Research: Planets, 114, E01003. Bockheim, J. G., I. B. Campbell, and M. McLeod (2007). Permafrost distribution and active‐layer depths in the McMurdo Dry valleys, Antarctica. Permafrost and Periglacial Processes, 18(3), 217-227. Brock, B. W., I. C. Willis, and M. J. Sharp (2006), Measurement and parameterization of aerodynamic roughness length variations at Haut Glacier d’Arolla, Switzerland, Journal of Glaciology, 52 (177), 281–297. Carr, M. H., and J. W. Head (2015). Martian surface/near‐surface water inventory: Sources, sinks, and changes with time. Geophysical Research Letters, 42(3), 726- 732. 19 Cassanelli, J. P., and J. W. Head III (2015), Firn densification in a Late Noachian 'icy highlands" Mars: Implications for ice sheet evolution and thermal response, Icarus, 253, 243-255. Conovitz, P. A., L. H. MacDonald, and D. M. McKnight (2006). Spatial and temporal active layer dynamics along three glacial meltwater streams in the McMurdo Dry Valleys, Antarctica. Arctic, Antarctic, and Alpine Research, 38(1), 42-53. Craddock, R. A., and A. D. Howard (2002), The case for rainfall on a warm, wet early Mars, Journal of Geophysical Research: Planets, 107, E11, 5111. Dunne, T., A. G. Price, and S. C. Colbeck (1976), The Generation of Runoff From Subarctic Snowpacks, Water Resources Research, 12(4), 677–685. Fassett, C. I., and J. W. Head (2008). The timing of martian valley network activity: Constraints from buffered crater counting. Icarus, 195(1), 61-89. Fastook, J. L., and J. W. Head (2015). Glaciation in the Late Noachian Icy Highlands: Ice accumulation, distribution, flow rates, basal melting, and top-down melting rates and patterns. Planetary and Space Science, 106, 82-98. Forget, F., R. Wordsworth, E. Millour, J.-B. Madeleine, L. Kerber, J. Leconte, E. Marcq, and R.M. Haberle (2013), 3D modelling of the early martian climate under a denser CO2 atmosphere: Temperatures and CO2 ice clouds, Icarus, 222 (1), 81– 99. Grant, J., R. P. Irwin, S. Wilson, D. Buczkowski, and K. Siebach (2011), A lake in Uzboi Vallis and implications for Late Noachian–Early Hesperian climate on Mars, Icarus, 212(1), 110-122. 20 Haberle, R. M. (1998), Early Mars Climate Models, Journal of Geophysical Research: Planets, 103(E12), 28467-28479. Head, J. W., and D. R. Marchant (2014). The climate history of early Mars: insights from the Antarctic McMurdo Dry Valleys hydrologic system. Antarctic Science, 26(06), 774-800. Heavens, N. G., M. I. Richardson, and A. D. Toigo (2008), Two aerodynamic roughness maps derived from Mars Orbiter Laser Altimeter (MOLA) data and their effects on boundary layer properties in a Mars general circulation model (GCM), Journal of Geophysical Research: Planets, 113, E02014. Hoke, M. R., and B. M. Hynek (2009), Roaming zones of precipitation on ancient Mars as recorded in valley networks, Journal of Geophysical Research: Planets, 114(E8). Hoke, M. R. T., B. M. Hynek, and G. E. Tucker (2011), Formation timescales of large Martian valley networks, Earth and Planetary Science Letters, 312, 1–12. Howard, A. D., J. M. Moore, and R. P. Irwin (2005). An intense terminal epoch of widespread fluvial activity on early Mars: 1. Valley network incision and associated deposits. Journal of Geophysical Research: Planets, 110(E12). Hynek, B. M., and R. J. Phillips (2003), New data reveal mature, integrated drainage systems on Mars indicative of past precipitation, Geology, 31 (9), 757–760. Hynek, B. M., M. Beach, and M. R. Hoke (2010). Updated global map of Martian valley networks and implications for climate and hydrologic processes. Journal of Geophysical Research: Planets (1991–2012), 115(E9). 21 Irwin, R. P., R. A. Craddock, and A. D. Howard (2005), Interior channels in Martian valley networks: Discharge and runoff production, Geology, 33 (6), 489–492. Irwin, R. P., A. D. Howard, and R. A. Craddock (2008). Fluvial valley networks on Mars. River confluences, tributaries and the fluvial network, 419-451. Janowicz, J. R., N. Hedstrom, J. Pomeroy, R. Granger, and S. Carey (2004). Wolf Creek research basin water balance studies. In: Northern Research Basins Water Balance, D. Kane and D. Yang (eds). IAHS Publication No. 290, IAHS: Oxfordshire; 195–204. Judson, A., and N. Doesken, (2000). Density of freshly fallen snow in the central Rocky Mountains. Bulletin of the American Meteorological Society, 81(7), 1577-1587. Kerber, L., F. Forget, and R. Wordsworth (2014). Sulfur in the Early Martian Atmosphere Revisited: Experiments with a 3-D Global Climate Model. Presented at Mars Atmosphere: Modelling and Observation, 5th International Workshop. Levy, J. S., J. W. Head, and D. R. Marchant (2008). The role of thermal contraction crack polygons in cold-desert fluvial systems. Antarctic Science, 20(06), 565-579. Levy, J. S., A. G. Fountain, M. N. Gooseff, K. A. Welch, and W. B. Lyons (2011). Water tracks and permafrost in Taylor Valley, Antarctica: Extensive and shallow groundwater connectivity in a cold desert ecosystem. Geological Society of America Bulletin, 123(11-12), 2295-2311. Levy, J. (2015). A hydrological continuum in permafrost environments: The morphological signatures of melt-driven hydrology on Earth and Mars. Geomorphology, 240, 70-82. 22 Mahat V. and D. G. Tarboton (2014), Representation of canopy snow interception, unloading and melt in a parsimonious snowmelt model, Hydrological Processes, 28, 6320–6336. Piepgrass, M. V., E. P. Krider, and C. B. Moore (1982), Lightning and Surface Rainfall During Florida Thunderstorms, Journal of Geophysical Research: Oceans, 87, C13, 11,193–11,201. Puurveen, H., R. C. Izaurralde, D. S. Chanasyk, J. R Williams, and R. F. Grant (1997), Evaluation of EPIC’s snowmelt and water erosion submodels using data from the Peace River region of Alberta, Canadian Journal of Soil Science, 77(1), 41–50. Reynolds, J. F., P. R. Kemp, K. Ogle, and R. J. Fernández (2004), Modifying the ‘pulse– reserve’ paradigm for deserts of North America: precipitation pulses, soil water, and plant responses, Oecologia, 141, 194–210. Singh, V. P. (2001). Snow and glacier hydrology (Vol. 792367677). Springer Science & Business Media. Tarboton, D. G. (1994). Measurements and modeling of snow energy balance and sublimation from snow. Tarboton, D. G., and C. H. Luce (1996), Utah Energy Balance Snow Accumulation and Melt Model (UEB): Computer model technical description and users guide. Utah Water Research Laboratory Utah State University and USDA Forest Service Intermountain Research Station. 64 pp. Tian, F., M. W. Claire, J. D. Haqq-Misra, M. Smith, D. C. Crisp, D. Catling, K. Zahnle, and J. F. Kasting (2010), Photochemical and climate consequences of sulfur outgassing on early Mars, Earth and Planetary Science Letters, 295, 412–418. 23 von Paris, P., J. L. Grenfell, H. Rauer, and J. W. Stock (2013). N2-associated surface warming on early Mars. Planetary and Space Science, 82, 149-154. Warren, S. G. (1982), Optical Properties of Snow, Reviews of Geophysics and Space Physics, 20, 67–89. Williams, K. E., O.B. Toon, J.L. Heldmann, C. McKay, and M.T. Mellon (2008), Stability of mid-latitude snowpacks on Mars, Icarus, 196, 565–577. Williams, R. M., and R. J. Phillips (2001). Morphometric measurements of Martian valley networks from Mars Orbiter Laser Altimeter (MOLA) data. Journal of Geophysical Research: Planets, 106(E10), 23737-23751. Wilson, L., and J. W. Head (2009), Tephra deposition on glaciers and ice sheets on Mars: Influence on ice survival, debris content and flow behavior, Journal of Volcanology and Geothermal Research, 185, 290–297. Wordsworth, R., F. Forget, E. Millour, J.W. Head, J.-B. Madeleine, and B. Charnay (2013), Global modelling of the early martian climate under a denser CO2 atmosphere: Water cycle and ice evolution, Icarus, 222 (1), 1–19. Wordsworth, R. D., L. Kerber, R. T. Pierrehumbert, F. Forget and J. W. Head (2015), Comparison of “warm and wet” and “cold and icy” scenarios for early Mars in a 3-D climate model, Journal of Geophysical Research: Planets, 120, 1201–1219. Yen, Y. C. (1981). Review of thermal properties of snow, ice and sea ice (No. CRREL- 81-10). Cold Regions Research and Engineering Lab, Hanover, NH. 24 Table 1. Values adopted for snowmelt model parameters. Parameter Value adopted Source Comment Surface aerodynamic 2 × 10-3 m Heavens et al. Over the poles; consistent with terrestrial values roughness Sensitivity analysis: (2008) over unvegetated snow/ice 1 × 10-4 m – 1 × 10-2 m Snow density 450 kg m-3 Cassanelli and Approximate midpoint of range between fresh Head (2015) snow and dense firn (100 to 830 kg m-3, respectively) Liquid holding capacity of 5% Singh (2001) Range of terrestrial values is 5 – 15%; value is a snow Sensitivity analysis: function of ice layers, surface slope, etc. 5 – 15% Snow saturated hydraulic 30 m hr-1 Cassanelli and conductivity Head, in prep. Thermally active depth of 2.0 m Tarboton et al. Assuming ice near 0°C, and accounting for substrate (1994) oscillations on ~30-day timescales Snow emissivity 0.99 Warren (1982) Insensitive to other variables Snow albedo 0.9 - Reasonable range of terrestrial values; chose 0.9 Sensitivity analysis: 0.6 – 0.9 for final simulations to represent conservative results, but results are not highly sensitive. Thermal conductivity of 0.022 Yen (1981), Yen (1981) empirical equation for thermal fresh snow Judson and conductivity as function of density; mid-range of Doesken (2000) average density from Judson and Doesken (2001) Energy content initial -9.7 × 104 kg m-3 Tarboton and Equivalent to 20 m snowpack at 250°K average condition Luce (1996) temperature Snow water equivalent 20 m Cassanelli and Firn-to-ice transition approaches this depth after initial condition (m) Sensitivity analysis: 20 – 100 m Head (2015) several thousand years Fraction of surface melt 1.0 - Infiltration considered separately in this study that runs off 25 Table 2. Valley network names, locations, and calculated formative runoff. Name Latitude Longitude Calculated runoff Source -1 Evros Vallis 12°S 12°E 4 mm day Hoke et al. (2011) Licus Vallis 3°S 126°E 3 mm day-1 Irwin et al. (2005) -1 Paraná Valles 24.1°S 10.8°W 1.1 cm day Irwin et al. (2005) -1 Unnamed VN #1 0°N/S 23°E 5 cm day Hoke et al. (2011) Unnamed VN #2 6.6°S 134.7°E 5.5 cm day-1 Irwin et al. (2005) 26 Table 3. Annual average temperature at studied valley networks. P = 600 mb, o = 25° = 2.5⋅10-5 = 5⋅10-5 = 1⋅10-4 = 2⋅10-4 Evros Vallis 239.5 °K 246.8 °K 257.4 °K 274.6 °K Licus Vallis 232.1 °K 239.0 °K 250.9 °K 268.1 °K Paraná Valles 240.0 °K 247.7 °K 259.7 °K 279.4 °K Unnamed VN #1 237.9 °K 245.4 °K 255.4 °K 273.2 °K Unnamed VN #2 237.2 °K 244.5 °K 257.8 °K 277.8 °K P = 600 mb, o = 41.8° = 2.5⋅10-5 = 5⋅10-5 = 1⋅10-4 = 2⋅10-4 Evros Vallis 233.0 °K 240.0 °K 251.7 °K 269.3 °K Licus Vallis 226.0 °K 232.8 °K 245.2 °K 263.8 °K Paraná Valles 234.2 °K 241.5 °K 253.7 °K 273.0 °K Unnamed VN #1 231.2 °K 238.9 °K 250.8 °K 269.8 °K Unnamed VN #2 230.4 °K 237.7 °K 250.9 °K 272.2 °K P = 1000 mb, o = 25° κ = 2.5⋅10-5 κ = 5⋅10-5 κ = 1⋅10-4 κ = 2⋅10-4 Evros Vallis 261.2 °K 267.2 °K 276.8 °K 286.0 °K Licus Vallis 254.1 °K 260.1 °K 269.3 °K 272.7 °K Paraná Valles 263.4 °K 270.5 °K 280.7 °K 290.0 °K Unnamed VN #1 259.7 °K 265.9 °K 275.0 °K 284.2 °K Unnamed VN #2 260.4 °K 267.5 °K 278.1 °K 286.5 °K P = 1000 mb, o = 41.8° = 2.5⋅10-5 = 5⋅10-5 = 1⋅10-4 = 2⋅10-4 Evros Vallis 246.8 °K 255.8 °K 270.4 °K 285.1 °K Licus Vallis 241.2 °K 249.4 °K 263.8 °K 273.2 °K Paraná Valles 249.6 °K 258.2 °K 274.1 °K 288.7 °K Unnamed VN #1 245.7 °K 254.1 °K 268.8 °K 283.8 °K Unnamed VN #2 245.4 °K 255.0 °K 271.3 °K 286.2 °K 27 Table 4. Annual cumulative rainfall at studied valley networks. P = 600 mb, o = 25° = 2.5⋅10-5 = 5⋅10-5 = 1⋅10-4 = 2⋅10-4 Evros Vallis 0m 0m 0m 0.118 m Licus Vallis 0m 0m 0m 0m Paraná Valles 0m 0m 0m 0.025 m Unnamed VN #1 0m 0m 0m 0.024 m Unnamed VN #2 0m 0m 0m 0.016 m P = 600 mb, o = 41.8° = 2.5⋅10-5 = 5⋅10-5 = 1⋅10-4 = 2⋅10-4 Evros Vallis 0m 0m 0m 0.003 m Licus Vallis 0m 0m 0m 0m Paraná Valles 0m 0m 0m 0.014 m Unnamed VN #1 0m 0m 0m 0m Unnamed VN #2 0m 0m 0m 0m P = 1000 mb, o = 25° = 2.5⋅10-5 = 5⋅10-5 = 1⋅10-4 = 2⋅10-4 Evros Vallis 0m 0m 0.215 m 0.428 m Licus Vallis 0m 0m 0m 0.163 m Paraná Valles 0m 0m 0.015 m 0.097 m Unnamed VN #1 0m 0m 0.215 m 7.99 mm Unnamed VN #2 0m 0m 1.16 mm 0.029 m P = 1000 mb, o = 41.8° = 2.5⋅10-5 = 5⋅10-5 = 1⋅10-4 = 2⋅10-4 Evros Vallis 0m 0m 0m 0.133 m Licus Vallis 0m 0m 0m 0.383 m Paraná Valles 0m 0m 0.025 m 0.093 m Unnamed VN #1 0m 0m 0m 0.022 m Unnamed VN #2 0m 0m 0m 0.152 m 28 Figure 1. Example of an interior channel in part of Licus Vallis (cf. Irwin et al., 2005), highlighted with white arrows. HRSC image mosaic. 29 Figure 2. Sensitivity of κ = 2.0 × 10-4 m2 kg−1, 25° obliquity, 1000 mb CO2 results at Paraná Valles to (a) albedo varying from 0.6 (blue curve) to 0.9 (green curve), (b) snow density varying from 100 (blue curve) to 800 (green curve) kg m-3, (c) snowpack surface roughness length varying from 0.0001 (blue curve) to 0.01 m (green curve), and (d) snowpack depth varying from 20 (green curve) to 100 (blue curve) m. a b 30 c d 31 Figure 3. Map of annual average surface temperature across Mars (shaded, °K) with spin-axis obliquity 25°, 1000 mb CO2, and grey gas κ = 2.5 × 10-5 (a), 5.0 × 10-5 (b), 1.0 × 10-4 (c), or 2.0 × 10-4 m2 kg−1 (d). a b 32 c d 33 Figure 4. Rolling 24-hour summed snowmelt rates (in m day-1) for Evros Vallis, with 600 mb CO2 and 25° spin-axis obliquity (a), 600 mb CO2 and 41.8° spin-axis obliquity (b), 1000 mb CO2 and 25° spin-axis obliquity (c), or 1000 mb CO2 and 41.8° spin-axis obliquity (d). Where melt is non-zero, melt for simulations with κ = 2.5⋅10-5 m2 kg−1 is plotted in purple, κ = 5⋅10-5 m2 kg−1 is plotted in blue, κ = 1⋅10-4 m2 kg−1 is plotted in green, and melt for simulations with κ = 2⋅10-4 m2 kg−1 is plotted in red. a 34 b c 35 d 36 Figure 5. Rolling 24-hour summed snowmelt rates (in m day-1) for Licus Vallis, with 600 mb CO2 and 25° spin-axis obliquity (a), 600 mb CO2 and 41.8° spin-axis obliquity (b), 1000 mb CO2 and 25° spin-axis obliquity (c), or 1000 mb CO2 and 41.8° spin-axis obliquity (d). Where melt is non-zero, melt for simulations with κ = 2.5⋅10-5 m2 kg−1 is plotted in purple, κ = 5⋅10-5 m2 kg−1 is plotted in blue, κ = 1⋅10-4 m2 kg−1 is plotted in green, and melt for simulations with κ = 2⋅10-4 m2 kg−1 is plotted in red. a 37 b c 38 d 39 Figure 6. Rolling 24-hour summed snowmelt rates (in m day-1) for Paraná Vallis, with 600 mb CO2 and 25° spin-axis obliquity (a), 600 mb CO2 and 41.8° spin-axis obliquity (b), 1000 mb CO2 and 25° spin-axis obliquity (c), or 1000 mb CO2 and 41.8° spin-axis obliquity (d). Where melt is non-zero, melt for simulations with κ = 2.5⋅10-5 m2 kg−1 is plotted in purple, κ = 5⋅10-5 m2 kg−1 is plotted in blue, κ = 1⋅10-4 m2 kg−1 is plotted in green, and melt for simulations with κ = 2⋅10-4 m2 kg−1 is plotted in red. a 40 b c 41 d 42 Figure 7. Rolling 24-hour summed snowmelt rates (in m day-1) for an unnamed valley network at 0°N, 23°E (here, "Unnamed VN #1"), with 600 mb CO2 and 25° spin-axis obliquity (a), 600 mb CO2 and 41.8° spin-axis obliquity (b), 1000 mb CO2 and 25° spin- axis obliquity (c), or 1000 mb CO2 and 41.8° spin-axis obliquity (d). Where melt is non- zero, melt for simulations with κ = 2.5⋅10-5 m2 kg−1 is plotted in purple, κ = 5⋅10-5 m2 kg−1 is plotted in blue, κ = 1⋅10-4 m2 kg−1 is plotted in green, and melt for simulations with κ = 2⋅10-4 m2 kg−1 is plotted in red. a 43 b c 44 d 45 Figure 8. Rolling 24-hour summed snowmelt rates (in m day-1) for an unnamed valley network at 6.6°S, 135°E (here, "Unnamed VN #2"), with 600 mb CO2 and 25° spin-axis obliquity (a), 600 mb CO2 and 41.8° spin-axis obliquity (b), 1000 mb CO2 and 25° spin- axis obliquity (c), or 1000 mb CO2 and 41.8° spin-axis obliquity (d). Where melt is non- zero, melt for simulations with κ = 2.5⋅10-5 m2 kg−1 is plotted in purple, κ = 5⋅10-5 m2 kg−1 is plotted in blue, κ = 1⋅10-4 m2 kg−1 is plotted in green, and melt for simulations with κ = 2⋅10-4 m2 kg−1 is plotted in red. a 46 b c 47 d 48 Figure 9. Annual average temperatures for GCM simulations with 1000 mb CO2, 25° spin-axis obliquity, and grey gas constants κ = 2.5⋅10-5 m2 kg−1 (a), κ = 5⋅10-5 m2 kg−1 (b), κ = 1⋅10-4 m2 kg−1 (c), and κ = 2⋅10-4 m2 kg−1 (d). a b 49 c d 50 Figure 10. Snowmelt rates at each study site in the 25°, 1000 mb GCM simulations, with rolling 24-hour summed rainfall rates (in m day-1) added in thick grey (κ = 1⋅10-4 m2 kg−1) and black (κ = 2⋅10-4 m2 kg−1) lines. a b 51 c d 52 e 53 Figure 11. Map showing annual cumulative rainfall (in m year-1) at all study sites with 1000 mb CO2, and grey gas κ = 5⋅10-5 m2 kg−1 (a), κ = 1⋅10-4 m2 kg−1 (b), and κ = 2⋅10-4 m2 kg−1 (c). Simulations with spin-axis obliquity 25° are shown at left and those with spin-axis obliquity 41.8° are shown at right. a b c 54 Figure 12. Map showing surface water ice distribution (in kg m-2) after ~400 model years of ice evolution with a 1000 mb pure CO2 atmosphere and spin-axis obliquity 25° (a) or 41.8° (b). In the Late Noachian Icy Highlands model, this represents the distribution of snow and ice that would have been available for melt upon warming episodes. a 55 b 56 Chapter 2 The Dorsa Argentea Formation and the Noachian-Hesperian Climate Transition K. E. Scanlon,1 J. W. Head,1 and J. L. Fastook2 1 Department of Earth, Environmental and Planetary Sciences, Brown University 324 Brook Street, Box 1846, Providence, RI, 02912, USA 2 School of Computing & Information Science, University of Maine 348 Boardman Hall, Orono, ME 04469 57 Abstract The Dorsa Argentea Formation (DAF), a set of geomorphologic units covering ~1.5 million square kilometers in the south circumpolar region of Mars, has been interpreted as the remnants of a large south polar ice sheet that formed near the Noachian-Hesperian boundary and receded in the early Hesperian. Determining the extent and thermal regime of the DAF ice sheet, as well as the mechanism and timing of its recession, can therefore provide insight into the ancient martian climate and the timing of the transition from a comparatively thick CO2 atmosphere to the present climate. We used the Laboratoire de Météorologie Dynamique (LMD) early Mars global climate model (GCM) and the University of Maine Ice Sheet Model (UMISM) glacial flow model to constrain climates allowing (1) development of a south polar ice sheet of DAF- like size and shape, (2) basal melting of this ice sheet in amounts and locations consistent with observed glaciofluvial landforms, and (3) recession of this ice sheet consistent with the preservation and crater exposure ages of landforms in the DAF. A large, asymmetric south polar ice sheet is a robust feature of GCM simulations with a 600 – 1000 mb CO2 atmosphere, with the shape resulting from the large-scale south polar topography of Mars and the strong dependence of surface temperature upon altitude in a thicker atmosphere. The extent of the modeled DAF ice sheet in UMISM simulations most closely matches that of the DAF when the water ice inventory of Mars is a ~137 m global equivalent layer (GEL) and the combined effects of a thicker CO2 atmosphere and an idealized greenhouse gas warm the surface by ~75 degrees near the poles. Basal and top-down melting occur in the locations where eskers are currently observed in artificially warmed 58 climates. In climates warmed only by CO2, however, top-down melting is rare, basal melting does not occur, and the warmest basal temperatures are towards the center of the ice sheet rather than in the distal locations where eskers are observed. Previously published crater exposure ages of eskers in the DAF indicate that eskers were being exposed as activity was ceasing in the equatorial valley networks, suggesting that the episodes of warming that allowed basal melting at the edges of the DAF ice sheet were the same episodes in which the valley networks were carved. 1. Introduction The Dorsa Argentea Formation (DAF), a set of geomorphologic units (Figure 1) covering ~1.5 million square kilometers surrounding the south pole of Mars (Hdl, Hdu, and HNu of Tanaka and Scott, 1987; Hdr, Hdv, Had, Hdd, Hdp, Hds, Hdc, and ANdc of Tanaka and Kolb, 2001; Hp, Hpe, and Ap of Tanaka et al., 2014), has been interpreted as the remnants of a larger south polar ice sheet that formed near the Noachian-Hesperian boundary and receded in the early Hesperian era (Head and Pratt, 2001; Fastook et al., 2012). Since it is high in altitude as well as latitude, the south pole of Mars would have been a much more effective cold trap than the north pole if the martian atmosphere was less tenuous in its early history (e.g. Wordsworth et al., 2013; Forget et al., 2013; Mischna et al., 2013; Urata and Toon, 2013). This effect, along with a larger surface water inventory earlier in martian history (e.g. Carr and Head, 2015), may account for the much greater size of the DAF relative to the present-day south polar ice deposits. Determining the extent and thermal regime of the DAF ice sheet, as well as the mechanism and timing of its recession, can therefore provide valuable insight into the 59 ancient martian climate and the timing of the transition from a thicker CO2 atmosphere to the present-day martian climate regime. In this work, we use early Mars global climate model (GCM) and glacial flow model simulations in order to constrain which of the possible late Noachian – early Hesperian climates could allow (1) the development of a south circumpolar ice sheet of similar size and shape to the present-day DAF, (2) basal melting of this ice sheet in amounts and locations consistent with the size and distribution of glaciofluvial landforms, and (3) recession of this ice sheet consistent with the preservation state and crater exposure ages of landforms in the DAF. Most studies of the DAF since the Mars Global Surveyor mission have supported the hypothesis (Allen, 1979; Howard, 1981; Kargel and Strom, 1992) that many features in the DAF are glacial in origin. Specifically, Head and Pratt (2001) argued that the DAF as a whole comprises the volatile-rich remnants of a large south polar ice sheet, whose retreat formed the surrounding fluvial features. The lobate contacts between the DAF plains and other units, and the position of fluvial channel heads near the edges of the DAF, suggest that the boundaries of the modern DAF correspond approximately to the boundaries of the former ice sheet. The ice sheet interpretation is based on the geomorphology of several landform classes. First, slightly sinuous ridges tens of meters high and kilometers to hundreds of kilometers long cluster in several regions of the DAF. Howard (1981) first suggested that these ridges were eskers, sedimentary deposits built in subglacial channels and left behind as ridges when the surrounding ice receded (e.g. Brennand, 2000). Topographic data from the Mars Orbital Laser Altimeter (MOLA) supported this hypothesis: Head and Hallet (2001a, 2001b) and Head and Pratt (2001) argued that the Dorsa Argentea and 60 other sinuous ridges in the DAF are eskers because (1) they are similar in scale, spacing, cross-sectional shape, sinuosity, branching patterns, and branching angles to terrestrial eskers; (2) unlike rivers, they do not uniformly follow topographic slopes; rather, like eskers, they change in scale and cross-sectional profile in response to underlying topography; and (3) their superposition relationships are consistent with their having formed subglacially and not subaerially. Due to the subglacial formation process of eskers, the age of the eskers themselves cannot be determined from crater counting; instead, the crater count on a population of eskers is a function of the time since the ice overlying the eskers was removed. Kress and Head (2015) used buffered crater counts to determine the exposure ages of five of the seven visible populations of eskers within the DAF. They found exposure ages ranging from 3.41 to 3.83 billion years, with the esker populations near the 270°W meridian having been exposed ~300 million years longer than the populations near the 90°W meridian. The Sisyphi Montes, a chain of mountains in the 0°W lobe of the DAF, present further evidence for past glacial conditions in the DAF. Murray et al. (1972) were the first to observe these mountains, and suggested a possible volcanic origin for them. On the basis of the morphology, morphometry, and distribution of the Sisyphi Montes, Ghatan and Head (2002) argued that the Sisyphi Montes are subglacially erupted volcanic edifices, rather than remnant crater rims or crater central peaks, as other workers had suggested in addition to the volcanic hypothesis (Murray et al., 1972; Peterson, 1977; Condit and Soderblom, 1978; Scott and Carr, 1978). Ghatan et al. (2003) interpreted the Cavi Angusti to have been formed due to localized ice melting caused by similar subglacial volcanic eruptions, as well as intrusive magmatism. Compact Reconnaissance 61 Imaging Spectrometer for Mars (CRISM) detections of hydrated sulfate spectra in the south circumpolar region are strongest in the Sisyphi Montes volcanic edifices, which may suggest that hydrothermal alteration occurred in the ancient glaciovolcanic environment (Wray et al. 2009, Ackiss and Wray 2014). Finally, fluvial features surrounding the deposit also point to a glacial origin. Milkovich et al. (2002) found that several impact craters on the eastern edge of the 0°W lobe of the DAF, including Main Crater, are connected by channels. The elevations of these channels indicate that at least 1000 km3 of water ponded within this system, and the location of these channels far from other DAF fluvial features suggests that the process that caused melting of the DAF ice sheet was not localized to one area. Ghatan and Head (2004) mapped five valleys emerging from across the DAF, which lack preserved dendritic tributaries, and three of which showed pitting in their source regions, consistent with the valleys representing drainage conduits during melting of the DAF. Dickson and Head (2006) found geomorphological evidence that pitted terrain and linear valleys at the edge of the DAF near Schmidt Crater formed as a kettled outwash plain and groundwater sapping valleys, respectively, and that this edge of the DAF ice sheet was in contact with a proglacial lake. The above described features all suggest that the Dorsa Argentea Formation was formed during the ablation of an areally extensive ice sheet, but neither large-scale ice melt nor the formation of a large south polar ice sheet are favored by the current martian climate. How did the climate and geologic processes that shaped the DAF ice sheet differ from those operating on present-day Mars? Any model explaining the genesis of the Dorsa Argentea Formation must explain the following key observations: 62 1. The Dorsa Argentea Formation (with an area of ~1.5 million km2) is more than an order of magnitude larger in area than the present-day southern residual ice cap (~100,000 km2). This difference could be caused by a larger total ice inventory at the Noachian-Hesperian boundary, or by climate conditions strongly favoring deposition at the south pole at that time, or both. 2. The areal centroid of the Dorsa Argentea Formation is offset from the present-day spin axis by ~8 degrees of latitude. This offset has been attributed to true polar wander by some workers (Kite et al., 2009). However, even in the tenuous modern martian atmosphere, climate dynamics cause asymmetric CO2 frost emplacement patterns about the south pole (with a persistent region of atmospheric supersaturation centered ~10° north of the pole between the 90°W and 45°W meridians; Colaprete et al., 2005), and a significantly warmer and/or thicker atmosphere at the Noachian- Hesperian boundary may have even more strongly influenced deposition and/or preservation of water ice. 3. The planform of the Dorsa Argentea Formation is not symmetric about the south pole, and instead has prominent lobes stretching down the 0°W and 90°W meridians. This may have been caused by asymmetric erosion, by multiple true polar wander events, or, as above, by asymmetric precipitation and temperature patterns. 4. Several populations of large and well-preserved eskers, which are morphometrically uniform and continuous over long distances, are present in the deposit. Wet-based conditions are necessary to create eskers, but any subsequent episodes of wet-based glacial sliding in these areas would have degraded or destroyed them (e.g. Kleman, 1994). 63 5. These eskers are densest near the eastern and western edges of the 90°W lobe and in Parva Planum. This could be explained by sediment supply being insufficient elsewhere in the deposit (such that subglacial tunnels may have existed elsewhere but no inverted sedimentary deposits were left), by basal melt being particularly abundant in these regions (due to thicker ice or warmer atmospheric temperatures), or by the eskers having formed beneath the margins of an ice sheet as it retreated past these regions. 6. Glaciovolcanic features (namely, tuyas and peripheral Hesperian volcanic flows with steep margins where they intersect the DAF) exist in the deposit, but none are directly associated with fluvial channels, most are not directly associated with eskers, and most eskers are not directly associated with glaciovolcanic features. The lack of eskers and channels surrounding most glaciovolcanic features could be explained by (a) inappropriate subglacial conditions for channel erosion when and where the eruptions occurred (e.g. low flow speed and a bedrock substrate), (b) insufficient debris supply from the eruptions and surrounding ice to create eskers (e.g. if the eruptions were primarily effusive and surrounding ice was relatively clean, little coarse-grained sediment would be carried in subglacial channels, making esker construction unlikely), and/or (c) by subsequent glacial erosion of any channels or eskers that formed (e.g., if the Sisyphi Montes significantly predate the rest of the deposit, later wet-based glacial sliding might have bulldozed most eskers or channels associated with the eruptions while leaving the edifices themselves relatively intact). The general lack of association between the observed eskers and subglacial edifices 64 likely suggests that the melting recorded by the majority of eskers in the DAF is not related to the construction of these edifices. 7. Some craters that lie along the boundaries of the DAF are heavily degraded on the side interior to the DAF but undisturbed on the side outside the DAF. This could be due to erosion by glacial sliding parallel to the DAF boundaries, or by the presence of an erodible substrate (such as ice) within the DAF at the time of impact which has since been removed. 8. Flat-topped glaciovolcanic features extending along the 0°W lobe of the deposit, interpreted as tuyas, stand up to ~1.8 km high. This indicates that ice in this region at the time of the eruptions was at least this thick (Ghatan and Head, 2002). 9. Some craters outside the continuous boundaries of the DAF are filled by isolated patches of the plains material with lobate margins that comprises much of the DAF, but the rims of the craters are not breached. This suggests downwasting of a formerly thicker unit to form the plains. To explore the environmental implications of these observations, we propose the following framework of hypotheses (Figure 2): (1) High PCO2 in the Late Noachian - Early Hesperian atmosphere increases the capacity of the atmosphere to communicate its heat to the surface, causing the surface to be warmer at lower altitudes, and hence favoring the buildup of a large ice sheet in the south polar highlands (Forget et al., 2013; Wordsworth et al., 2013). The probability density of the spin-axis obliquity of Mars throughout its history peaks at 41.8° (Laskar et al., 2004); low-obliquity excursions may have further enhanced migration of ice to the poles. (2) Either unusual atmospheric 65 warming (e.g. by punctuated volcanic release of greenhouse gases or by an orbital configuration favoring warm south polar summers), the onset of subglacial volcanic activity, or buildup of the ice to sufficient thickness allows basal and/or top-down melting to occur beneath the parts of the ice sheet where the Dorsa Argentea and other esker populations are observed. (3) Any volcanism or any mechanism causing unusual atmospheric warming ceases (such that conditions for basal melting are no longer met regardless of ice thickness), or ice becomes less stable at the south pole (due to high spin- axis obliquity excursions, depletion of total atmospheric pressure such that the influence of surface elevation on ground surface temperature diminishes and ice accumulation at the south pole is no longer as strongly favored, or a decrease in the total surface water budget). Further recession of the ice sheet is therefore entirely cold-based, allowing (4) the preservation of several populations of eskers observed in the present day DAF. To distinguish between these hypotheses, we first use the morphometry of the eskers in the Dorsa Argentea Formation to quantitatively estimate the meltwater flux through the subglacial channels they formerly occupied. We then conduct a series of Early Mars simulations with the Laboratoire de Météorologie Dynamique (LMD) GCM (Forget et al., 2013; Wordsworth et al., 2013), varying atmospheric pressure, degree of atmospheric warming, spin-axis obliquity, and initial ice distribution in order to determine what climate configurations could cause the buildup and decay of a large south circumpolar ice sheet consistent with geological observations. We used climate data from these GCM simulations as input to the Mars-adapted University of Maine Ice Sheet Model (UMISM; Fastook et al., 2008; 2012; 2015) in order to simulate the ice sheet profile, and the rate and locations of basal melting, under these conditions. 66 2. Estimates of subglacial melt fluxes and melt longevity from esker morphometry To constrain the amount of melting at the base of the DAF ice sheet (or infiltrated to the base through englacial fractures or moulins), leading to the formation of the eskers, we follow the methodology of Banks et al. (2009) in calculating the flux through the subglacial channels that became the eskers. Specifically, velocity is calculated from a form of the Darcy-Weisbach equation appropriate for channel-full subglacial flow (Komar, 1979; Lorang and Hauer, 2003; Banks et al., 2009): 8 != !"# !! where g is gravitational acceleration; R is the hydraulic radius of the channel, obtained by dividing its cross-sectional area by its perimeter; S is the channel bottom slope (see Lorang and Hauer, 2003); and the dimensionless Darcy-Weisbach friction coefficient fR is given by ! !! !.! = 2 log ℎ/!!" + 1 (Leopold et al., 1964). In this equation, h is channel flow depth and D84 is the 84th percentile grain diameter of sediments in the channel bed (a common metric for describing typical coarse-grained sediments in a sample). Discharge Q through the former subglacial channel can then be calculated by measuring the esker width and assuming a cross-sectional shape for the channel. We also followed Banks et al. (2009) in (1) choosing a flow depth range of 1 - 10 m, bracketing the range of flow depths expected from terrestrial analogy; (2) choosing an esker sediment size range of 0.5 - 64 mm, i.e. 67 from medium sand to coarse gravel, the size range observed in terrestrial eskers; and (3) approximating the tunnel shape as a broad triangle in order to calculate its cross-sectional area and hence flux through the tunnel. We chose to perform this analysis only on esker systems within the DAF that are well-preserved (minimizing the possibility of error in esker width) and not mantled by debris (minimizing the error in channel-bottom slope). Two of the seven systems of eskers in the DAF fit these criteria: those in Parva Planum, and the Dorsa Argentea. Kress and Head (2015) measured a characteristic width of 1500 m for main-trunk eskers in Parva Planum and 2000 m for those among the Dorsa Argentea. We measured slopes over the length of the entire esker system from the MOLA 512 pixels per degree gridded topography dataset (Smith et al., 2003). The resulting discharges (Table 1) are in the range 6.5 × 102 – 6.5 × 104 m3s-1 for Dorsa Argentea and 5.6 × 102 – 5.6 × 104 m3s-1 for Parva Planum, comparable to the results Banks et al. (2009) found for the ridges interpreted as eskers in the Argyre basin (~1 × 104 m3s-1 on average). For terrestrial glaciers, peak seasonal subglacial discharges of the order 102 – 103 m3 s-1 are typical (e.g. Ensminger et al., 1999; Wadham et al., 2001; Flowers et al., 2003; Bingham et al., 2005; Lappegard et al., 2006; Bartholomaus et al., 2011), and peak glaciovolcanic jökulhlaup discharges can reach 105 – 106 m3 s-1 (e.g. Tómasson, 1996, 2002; Maizels, 1997; Baker, 2009). The range of possible formative discharges for the DAF eskers falls within the range of those calculated from terrestrial eskers (Cummings et al., 2011). The source of most sediment in terrestrial eskers is debris melted out from the surrounding basal ice (e.g. Hooke and Fastook, 2007). If this was also true for the DAF 68 eskers, an order of magnitude minimum construction time can be estimated from the scale of the eskers and the calculated fluxes through them (Table 2). Typical dimensions for well-preserved eskers in the Dorsa Argentea are approximately 2 km wide, 80 m high, and eskers are continuous for up to ~400 km. Approximating the esker cross-section as rectangular, for basal debris contents ranging from 1 – 10% by volume, 6.4 × 1011 – 6.4 × 1012 m3 of ice would need to be melted to construct such an esker. At discharges ranging from 6.5 × 102 – 6.5 × 104 m3s-1, this amount of water would take between ~114 days and ~320 years to move through the tunnels. For the Parva Planum eskers, typically 1.5 km wide, 50 m high, and continuous for as long as ~470 km, the estimated minimum construction time range is between ~75 days and ~210 years for the same range of debris content. 3. Modeling the development and ablation of the DAF in an Early Mars GCM 3.1. Climate conditions under which a large, asymmetric ice sheet can form The Laboratoire de Météorologie Dynamique (LMD) “generic” GCM can be configured to simulate the climates and paleoclimates of a wide variety of solar system bodies and exoplanets, including a hypothesized Noachian-Hesperian Mars with a much denser CO2 atmosphere than at present (Forget et al., 2013; Wordsworth et al., 2013; Kerber et al., 2013; Kerber et al., 2015; Wordsworth et al., 2015; Bouley et al., 2015). To determine the climate conditions that favor the development of a DAF-like ice sheet, we ran the LMD GCM in early Mars mode with 64 grid points in longitude, 48 grid points in latitude, and 25 vertical levels. Atmospheric composition was pure CO2, and modern surface topography was used. To investigate the effect of pressure on the evolution of ice 69 at the south pole, simulations were initially conducted with atmospheric pressure of 600, 1000, and 1500 mb. Simulations conducted with pressure substantially higher than modern Mars but lower than 600 mb do not represent a stable climate state, as the martian atmosphere is unstable to CO2 condensation at the poles in this range (Wordsworth et al., 2015; see Soto et al., 2015, for similar findings using another Mars GCM), and were not used. Ice was initially present as a ~5 m thick layer covering the southern highlands (delineated for purposes of the GCM as all grid points above the +1 km datum); this unrealistically small ice inventory was chosen to enable ice to reach a steady state in a computationally reasonable time. After the third, fifth, seventh, and ninth model years, the net ice accumulation and ablation in that year is multiplied by 100 before the simulation continues (see Wordsworth et al. 2013); our ten-year simulations therefore correspond to approximately 400 years of ice evolution. All simulations were conducted at both 25° spin-axis obliquity, similar to modern Mars, and 41.8° spin-axis obliquity, which is the most probable value over martian history (Laskar, 2004), in order to determine whether low obliquity is required to build or maintain a large south polar ice sheet. Solar constant was set to 1024.5 W m-2 at 1 AU, consistent with a faint young sun (Gough, 1981), except for one simulation where it was set to 1761.9 W m-2 at 1 AU in order to investigate ice distribution in a much warmer climate (see Wordsworth et al., 2015). Mars was set in a circular orbit with radius 227.9 × 106 km. In their investigation of the climate conditions necessary to allow basal melting of the DAF ice sheet, Fastook et al. (2012) uniformly scaled the surface temperature field from Amazonian Mars GCM simulations (Madeleine et al., 2009), since early Mars GCM 70 data were not yet available at the time. Annual average surface temperature is approximately symmetric about the south pole on modern Mars (e.g. Kieffer 2013), reflecting the primary importance of surface radiation balance on planets with thin atmospheres (e.g. Read and Lewis, 2004). In early Mars simulations, however, with weaker insolation and a thicker atmosphere capable of communicating more heat to the surface, the annual average surface temperature shows an approximate wavenumber-3 pattern about the south pole, i.e. a pattern with cold lobes extending northward from the south pole along the 0°W, 90°W, and ~225°W meridians (Figures 3, 4). The influence of topography on annual average surface temperature becomes more obvious as surface pressure increases (Figures 3, 4). We performed several additional simulations at 1000 mb surface pressure and 25° obliquity to test the effect of major topographic features on the temperature pattern near the south pole. Flattening the Tharsis bulge removes the cold lobe of the surface temperature pattern that extends along the 90°W meridian, south of Tharsis (Figure 5a), leaving a surface temperature minimum elongated in the direction of the 0° - 180° meridians. Removing the Hellas and Argyre basins shrinks the cold lobe extending along the 0° meridian (Figure 5b), leaving a surface temperature minimum elongated primarily along the 90°W meridian. Completely flattening the planetary topography results in a surface temperature minimum symmetric about the south pole (Figure 5c). Together, the effects of surface pressure and topography suggest that the temperature pattern reflects the influence of both latitude-dependent insolation, which is more important in thinner atmospheres and atmospheres with less greenhouse warming, and the adiabatic cooling effect causing colder temperatures at more elevated surfaces, 71 which is more important in thicker atmospheres (Forget et al., 2013; Wordsworth et al., 2013). This effect is further illustrated by plotting annual average surface temperature in our simulations against surface elevation at each model grid point (Figure 6); the correlation (r2) between the two variables increases from 0.28 at 600 mb to 0.84 at 1500 mb for 25° spin-axis obliquity, and from 0.37 at 600 mb to 0.95 at 1500 mb for 41°8. (The poor correlation at altitudes near +1000 m in all simulations may reflect a greater importance of insolation geometry and dynamical effects, such as slope winds, along the steep slopes of the dichotomy boundary and Hellas basin.) In summary, the prominent 0°W and 90°W lobes of the DAF in planview suggest that the DAF ice sheet formed in a climate where elevation significantly affected surface temperature. A key conclusion Fastook et al. (2012) drew from their modeling results is that annual average surface temperatures in the DAF region at the Noachian-Hesperian boundary would need to be between ~198°K (if Noachian-Hesperian geothermal heat flux was 65 mW m-2) and ~223°K degrees (if it was 45 mW m-2) in order for a DAF-like ice sheet to undergo basal melting. When only CO2 greenhouse warming is present, annual average temperatures in the vicinity of the Dorsa Argentea eskers (e.g. Figure 7) exceed 198°K for all surface pressures studied, but exceed 223°K only in the 1500 mb simulations. Most estimates for martian surface pressure at the Noachian-Hesperian boundary are below 1500 mb; for example, Lammer et al. (2013) estimate that Mars could have accumulated an atmosphere of up to 1 bar CO2 from outgassing and impacts by the end of the Noachian, and Kite et al. (2014) constrained atmospheric pressure to below 0.9 ± 0.1 bar at 3.6 Ga from crater sizes in Aeolis Dorsa. If the early martian atmosphere was warmed only by CO2, a Noachian-Hesperian geothermal heat flux of 72 greater than 45 mW m-2 is therefore likely to be required to meet the basal melting criteria of Fastook et al. (2012). Top-down melting is not significant in our GCM simulations except in the simulations with artificially high solar flux. In our simulations at relatively low pressure (600 mb) and very high obliquity (60° or 70), the DAF ice sheet undergoes some surface melting, but only for a few days each year. Even if this melt were as rapid as esker geomorphometry requires, it would be unlikely to propagate from the surface to the base of the ice sheet in such a cold climate. In the 41.8° obliquity, 1500 mb surface pressure simulation, which has the highest global average temperature of our baseline runs, surface temperatures throughout the vast majority of the DAF region are never sufficient for even transient top-down melting (Figure 8). Ice accumulation also peaks in a wavenumber-3 pattern about the south pole (Figures 9, 10), with thin 90°W and 0°W lobes present at all pressures studied. When the simulated solar constant is increased to 1761.9 W m-2 at 1 AU (Figure 11), which warms Mars to the point where rainfall occurs in the equatorial highlands and surface temperatures are above freezing at the poles for much of the year, an ice cap with a similar extent and planform still accumulates at the south pole. The lobes are more prominent at higher obliquity and in the high solar flux scenario, likely due to warmer regional temperatures at the poles in either scenario; the highest annual snowfall rates and the coldest annual average temperatures both occur over the local topographic highs along the 0° and 90°W lobes. 3.2. Ablation pattern and rates for the DAF ice sheet 73 We have hypothesized that the recession of the DAF ice sheet may have been caused by (1) loss of the thicker Noachian-Hesperian atmosphere, causing the south pole to become a less favored site for ice deposition; (2) high-obliquity excursions, leading ice to become less stable at the poles; or (3) strong climate warming, possibly the same warming episodes that allowed the development of the equatorial valley networks, leading to meltback of the ice sheet. We investigated these possibilities, respectively, by initializing the GCM with ice as a uniform layer present only in grid points occupied by the present-day geologic units that comprise the DAF, and (1) setting a low atmospheric pressure (50 or 200 mb), (2) increasing obliquity (60° or 70°) with a 600 mb or 1000 mb CO2 atmosphere, or (3) markedly increasing solar flux (1761.9 W m-2 at 1 AU) with a 1000 mb CO2 atmosphere (see Wordsworth et al., 2015). Initial ice thickness was set to ~180 m, a factor of 10 thinner than the minimum expected for the DAF, in order to allow evolution of the ice sheet in a computationally reasonable time. As in the simulations of the construction of the DAF ice sheet, the LMD GCM was run at 64 x 48 horizontal grid point resolution with 25 vertical levels, with a pure CO2 atmosphere, modern surface topography, a circular orbit with radius 227.9 × 106 km, and 100x acceleration of ice accumulation/ablation to allow the removal of ice in a computationally reasonable time. While moderately high obliquity (41.8°) appears to facilitate the growth of a large south polar ice sheet with a wavenumber-3 pattern to the same extent as low (25°) obliquity, very high obliquity results in ablation (Figure 12). At either 60° or 70° spin- axis obliquity in a 1000 mb atmosphere, 1.5 × 104 kg m-2 of ice (or about 16 meters) is removed from the initial DAF-like ice sheet over the course of 10 model years, simulating ~400 years of ice evolution (Figure 12a,b). Assuming constant ablation rates, 74 a 2 km thick ice sheet could therefore be removed if high obliquity were maintained in a 1000 mb atmosphere for ~50 ky. At 600 mb surface pressure, removal rates are similar at 60° obliquity, but increase substantially at 70° obliquity (Figure 12c,d), reflecting the lower heat transport capacity of a thinner atmosphere and the greater importance of the peak insolation at the poles. The probability of Mars reaching 60° or 70° obliquity at any point in the past 5 Ga is ~95% or ~9%, respectively, and such excursions would have been rare (Laskar et al., 2004). In a 200 mb atmosphere (Figure 13), the DAF-like ice sheet sublimes slowly, at a rate of ~0.075 kg m-2 yr-1, or ~0.08 mm yr-1. At this rate, a 1.8 km thick ice sheet would have taken ~22.5 million years to recede. Reducing atmospheric pressure to 50 mb does not increase the rate of sublimation, despite the further decrease in the importance of altitude as a control on surface temperature. Instead, ice accumulates on the DAF-like ice sheet at a rate of up to ~0.8 mm yr-1. While the altitude of the southern highlands no longer contributes to the efficacy of the south polar region as a cold trap at this pressure, the high albedo of the ice sheet enables it to function as an efficient cold trap in the insolation-dominated climate. The role of albedo is illustrated by repeating the simulation with the albedo of ice set equal to the albedo of the ground (Figure 14); without the effect of high albedo, the mass balance of ice over the DAF ice sheet is again negative. As expected, the most efficient mechanism of removing the DAF ice sheet is warming the climate, accomplished here by increasing the solar flux. Ablation rates are on the order of 1 m yr-1 in this climate scenario. Using the 2 × 104 kg m-2 contour as a proxy for the edge of the ice sheet allows the pattern of recession to show clearly over smaller-scale variability (Figure 15). The distribution of ice after 20 model years in this 75 simulation is similar to the distribution after 20 model years in scenarios that were initialized with ice throughout the highlands, indicating that the predicted pattern of ice at the south pole is robust throughout our simulations and not sensitive to initial ice conditions. Notably, the regions of the initial ice sheet that are removed in the first 200 years of high flux simulations starting from a DAF-like ice sheet (regions between purple and red contours in Figure 15) correspond to the regions where esker populations have been exposed longest, as calculated from buffered crater counting by Kress and Head (2015). The Dorsa Argentea esker population, which has the next oldest exposure age, lies in a region of the ice sheet that is removed within the next 100 years of ablation (region between red and orange contours in Figure 15). The eskers with the oldest exposure ages occur in Promethei Basin, a topographic low; the decreased stability of ice at lower altitudes is likely to have contributed to the relatively early ice loss over this esker population. 4. Modeling the DAF ice sheet with a Mars-adapted glacial flow model Since simulating ice flow and basal ice temperature is beyond the scope of global climate models, our results from the GCM cannot fully determine whether the stable extent of the south polar ice sheet in a given climate matches that of the Dorsa Argentea Formation, or the amount and locations of basal melting. To address these questions, we used surface temperatures and ice accumulation and ablation rates from the early Mars GCM simulations as input to the University of Maine Ice Sheet Model (UMISM), adapted for Mars (Fastook et al., 2008; 2012; 2015). 76 Inputs to UMISM are (1) ice sheet bed topography, from MOLA; (2) annual average ground surface temperature and annual surface mass balance of water ice, from GCM simulations; and (3) Noachian-Hesperian geothermal heat flux, within the range of estimates from lithospheric thickness models (Montési and Zuber, 2003; Solomon et al., 2005). UMISM calculates the internal temperature structure (and hence the mechanical properties) of the ice from these data, using the shallow ice approximation; ice flows by internal deformation and, if basal melting occurs, sliding. Output includes the equilibrium ice sheet surface topography, amount of melt (m m-2) and rate (mm yr-1) of basal melting, basal ice temperatures, and ice flow velocity (mm yr-1). UMISM data can answer three key questions that the GCM simulations alone cannot: (1) Under which climate and geothermal heat flux conditions is the footprint of a modeled south polar ice sheet most similar to that of the DAF? GCM results delineate the zones where ice accumulates and ablates, but cannot account for flow speed and direction, and hence cannot determine the equilibrium shape and thickness of the ice sheet. Compare, for example, the GCM ice accumulation pattern in the Tharsis Montes region in Forget et al. (2006) to the footprint of the same ice sheets simulated using UMISM in Fastook et al. (2008); the observed glacial deposits (Head and Marchant, 2003), and those modeled by UMISM, extend hundreds of kilometers downslope of the region where ice accumulated in the GCM. (2) Where, and under which climate and geothermal heat flux conditions, does basal melting occur? Basal melting is not modeled in the GCM; its distribution and magnitude are influenced by ice thickness, surface temperature, and geothermal heat flux. (3) How much basal melt occurs? Can basal melt alone generate melt at the rate estimated from geomorphometry for the DAF eskers? 77 To answer these questions, we conducted simulations with UMISM, varying the following parameters: (1) GCM climate: 600 mb pure CO2; 1000 mb pure CO2; “warm, wet early Mars” achieved by adding an idealized greenhouse gas with wavelength- independent absorption coefficient κ = 2.5 × 10-5, 5 × 10-5, 1 × 10-4, 1.5 × 10-4, or 2 × 10-4 m2 kg−1 (i.e. “grey gas”); (2) global ice inventory: 20, 40, 80, or 200 million km3, i.e. a ~137 m, ~275 m, ~550 m, or ~1375 m global equivalent layer (GEL); and (3) geothermal heat flux: 45, 55, or 65 mW m-2. GCM data from the third year of model runs was used as input to UMISM, as surface temperatures have reached a stable state by this point. Global potential ice accumulation and ablation rates are determined by adding a uniform layer of ice to the planet and measuring the change over the course of the model year. Spin-axis obliquity was set to 41.8° for the GCM runs used with all UMISM simulations, as the final distribution of ice was not strongly dependent on obliquity in our GCM results. The modeled ice sheet surface (Figures 16, 17, 18) shows the same wavenumber- 3 pattern as the GCM snow accumulation, but with greater continuity and a somewhat greater spatial extent than in the GCM due to ice flow. Both the 90°W and 0°W lobes are much smaller than the observed DAF lobes in the 600 mb simulation (Figure 16). The extent of the ice sheet is somewhat greater in the 1000 mb simulation, though still not as large as that of the DAF. Since the distribution of ice at the south pole is similar in the 600 mb and 1000 mb climates (Figure 10), this difference is likely to reflect the influence of warmer atmospheric temperatures on glacial flow in the 1000 mb simulation (Figure 19). Increasing the global ice inventory (Figure 17) slightly increases the extent of the modeled DAF ice sheet, but the primary effect of increased ice inventory is to increase the extent, thickness, and continuity of ice sheets on the rims of the Hellas and Argyre 78 basins, and in the highland mid-latitudes from 90°W – 270°W. The effect of increased geothermal heat flux on the extent of the DAF (Figure 18) is negligible. No basal melting occurs in the UMISM simulations with temperature and ice mass balance data from either the 600 mb or 1000 mb GCM climates (Figures 20, 21, 22). For these climates, annual average basal temperatures approach the melting point only in simulations with ice inventories of ~550 m or ~1375 m GEL, which are likely to be unrealistically high values for the late Noachian - early Hesperian martian ice inventory (see reviews by Lammer et al., 2013; and Carr and Head, 2015). Furthermore, when geothermal heat flux or ice thickness is increased to the point where basal temperatures come within 10° – 20° of the melting point, the highest basal temperatures occur near the central, southernmost part of the ice sheet, rather than in the distal regions of the ice sheet (Figures 23, 24). GCM surface atmospheric temperatures in simulations with pure CO2 atmospheres do meet the criteria Fastook et al. (2012) presented for basal melting over the regions where eskers are observed, but in the absence of climate data for early Mars, Fastook et al. (2012) had assumed a radially symmetric pattern of mass accumulation extending from the south pole to the northern limits of the DAF, and spatial patterns of surface temperature and sublimation similar to those on modern Mars. As discussed in section 3, unlike on modern Mars, in a thicker atmosphere surface topography exerts a strong influence on surface temperature and mass balance, and ice in the GCM simulations is consequently either absent or insufficiently thick to allow basal melting in these regions. 79 In the UMISM simulations driven by a “cold and icy” GCM climate, where Mars is warmed only by CO2, the simulated DAF ice sheet is smaller than the DAF, and average basal temperatures never exceed freezing. Increasing global ice inventory or geothermal heat fluxes increases basal temperatures, but (1) no combination of realistic parameters is sufficient to cause any basal melting, let alone at the rates implied by the morphometry of eskers in the DAF, and (2) the warmest basal temperatures are near the center of the modeled DAF ice sheet in these simulations, rather than at its edges, closer to where eskers are observed. We therefore conclude that the basal melting of the DAF ice sheet could not have occurred to an extent or in a spatial pattern consistent with observations on an early Mars warmed only by CO2. Similar to the equatorial valley networks, which also formed at the Noachian-Hesperian boundary, the development of eskers and fluvial channels in the DAF would have required additional warming by an unknown mechanism. Since a physically plausible combination of greenhouse gases that could warm early Mars by more than a few degrees has not yet been identified, we added idealized “grey gases” of five different strengths (absorption coefficient κ = 2.5 × 10-5, 5 × 10-5, 1 × 10-4, 1.5 × 10-4, or 2 × 10-4 m2 kg−1) to our GCM simulations in order to investigate the shape and basal melt distribution of the DAF ice sheet in a warmer climate. The weakest grey gas raises annual average surface temperatures by 9 – 20°C relative to the scenario with a 1000 mb pure CO2 atmosphere; the strongest raises them by 40 – 70°C (Figure 25). A salient result (Figure 26) is that the distribution of ice in our most weakly warmed simulations, with κ = 2.5 × 10-5 or 5 × 10-5, is similar to that in the pure CO2 80 simulations, with most ice on the Hellas and Argyre basins and elsewhere in the highlands, and a south polar cap smaller than the DAF. In our most strongly warmed simulations, with κ = 1.5 × 10-4 or 2 × 10-4, most ice migrates to high topography in the mid-latitudes. In simulations with moderate warming, κ = 1 × 10-4, however, a south polar ice sheet develops with broad lobes extending to ~60°S along the 0°W and 90°W meridians. Rather than increasing the thickness of the ice sheet, increasing the global ice inventory in the κ = 1 × 10-4 simulations to 550 m GEL increases its extent past the observed boundaries of the DAF (Figure 27c). Decreasing the ice inventory to ~137 m GEL (Figure 27a) reduces the extent of the ice sheet near Argyre and Hellas basins, and is more consistent with observations than our nominal baseline inventory of 275 m GEL (Figure 27b). Basal melting rates in the κ = 1 × 10-4 simulations (Figures 28c, 29) are greatest at the edges of the ice sheet, consistent with the observed locations of eskers. Furthermore, the locations of maximum melting correspond approximately to the locations of the Dorsa Argentea, Promethei Basin, and Parva Planum esker populations, as well as the locations of the hypothesized open-basin lake systems near Argentea Planum and Main Crater (Figure 1). At least some top-down melting occurs throughout the ice sheet at κ = 1 × 10-4 simulations and higher. Our results indicate that a climate warmed by 38 – 48° near the poles (Figure 25c), with a global ice inventory of ~137 m, allows a stable ice sheet to develop at the south pole with a similar planform and extent to present-day planform and extent of the DAF, and to undergo basal melting, with maximum melt rates near the locations where 81 glaciofluvial landforms are observed. The 1500 m thickness contour of the modeled ice sheet in all κ = 1 × 10-4 scenarios studied extends past some, though not all, of the Sisyphi Montes. Since the heights of the Sisyphi Montes are thought to indicate a minimum ice thickness of 1800 m when they formed (Ghatan and Head, 2002), this suggests that a somewhat colder climate (limiting the flow of the ice sheet) with a greater ice inventory might more exactly match observations. However, the extent and the qualitative features of the deposit are closely matched by our κ = 1 × 10-4, ~137 m GEL scenario. 5. Discussion We have shown that if surface pressure was high enough on early Mars that surface temperatures depended substantially on surface altitude (Forget et al., 2013; Wordsworth et al., 2013), annual mean surface temperature patterns are modulated by both large-scale and local topography to create a cold trap whose wavenumber-three pattern about the south pole resembles the footprint of the Dorsa Argentea Formation. Therefore, the extent of the Dorsa Argentea Formation and associated units are likely to reflect the extent of the ice sheet that formed them, rather than asymmetric erosion of an approximately axisymmetric ice sheet. Kite et al. (2009) interpreted the offset of the DAF areal centroid from the south pole of Mars as the result of true polar wander, with the centroid of the DAF representing a southern paleopole dating to the Noachian-Hesperian boundary. In our GCM simulations, however, ice accumulation is centered several degrees north of the present- day rotational pole, along the same meridian as the areal centroid of the DAF (Figure 30). 82 Furthermore, our UMISM results from climates warmed by idealized greenhouse gases or artificially enhanced solar flux clearly illustrate that substantial ice flow in the direction of this offset is expected in warm climates (Figures 26, 29). We propose that the offset of the DAF areal centroid from the rotational south pole is due to downstream flow in a warmer climate after asymmetric ice accumulation in a thicker atmosphere, and that true polar wander need not be invoked to explain it. Recently, Bouley et al. (2015) conducted GCM simulations with the poles located near the present-day coordinates 100.5°W, 71.1 °N and 79.5 E, 71.1 °S, i.e. where the poles would be expected when Tharsis had not yet been constructed and the orientation of Mars about its spin-axis was controlled primarily by the topography dichotomy. They found that if this were the orientation of Mars at the time of valley network formation, the valley networks could have formed from precipitation as either rain or snow in a near- axisymmetric inter-tropical convergence zone. Bouley et al. (2015) note that these calculated paleopoles correspond with regions of low neutron counts, indicating that ground ice may be present and that polar ice caps may have covered them in the geologic past. The age of this ice cannot be determined, however; even if the inferred ice deposits do correspond to paleopoles, they may predate the valley networks. Crater counts (Fassett and Head, 2008; Kress and Head, 2015) indicate that the DAF eskers were being exposed to the surface during the same period of time that activity was ceasing in the valley networks. While crater ages alone cannot prove that the eskers did not form before the valley networks (e.g. if the eskers remained ice-covered for many years after their construction), they are inconsistent with the valley networks having formed prior to the DAF eskers. Furthermore, the lobe of cold temperatures and 83 ice accumulation extending northward along the 90°W meridian is not present in our GCM simulations without the Tharsis bulge (Figure 5a), which suggests that Tharsis was present by the time the DAF ice sheet formed. There is no evidence in our present- topography simulations that ice accumulation was concentrated in the hypothetical pre- Tharsis paleopoles during the period in which the DAF formed, nor is there evidence in the simulations of Bouley et al. (2015) for an extensive, asymmetric ice sheet near what is now the south pole. We therefore find that a scenario in which emplacement and melting of the DAF occurred simultaneously with the erosion of the valley networks, and after the emplacement of Tharsis, is more consistent with the globally integrated geologic evidence than the scenario of Bouley et al. (2015). We have presented three hypotheses for the formation of the eskers in the Dorsa Argentea Formation. First, the eskers may have formed in a cold martian climate (i.e. one in which temperatures do not approach 273°K near the equator) due to basal melting in the thickest regions of a polythermal ice sheet, either by “bottom-up” (Fastook et al., 2012) basal melting due to enhanced geothermal gradients (Figure 31a), bottom-up basal melting due to anomalously thick ice (Figure 31b), or “top-down” basal melting due to modestly increased atmospheric temperatures (Figure 31c). Second, the eskers may have formed in either a cold or warm climate due to subglacial volcanic activity, where either direct melting by extrusive volcanism or enhanced geothermal gradients due to intrusive volcanic activity allowed basal melting to occur underneath an otherwise cold-based DAF ice sheet (Figure 31d). Third, the eskers may have formed in a warm climate (i.e. one in which temperatures near the equator frequently exceed 273°K), by a combination of basal melting beneath the warmest parts of the ice sheet (Figure 31c) and infiltration of 84 surface melt to the base of a retreating ice sheet (Figure 31e). Retreat of the ice sheet, exposing the eskers, could have been driven by a global decrease in water inventory (decreasing the amount of water available to form a south polar ice sheet), a global decrease in surface pressure (decreasing the efficacy of the south pole in particular as a cold trap where ice remains stable), or extreme excursions in spin-axis obliquity (making surface ice less stable at either pole). Basal melting in a cold climate (Figures 31a, b, c) is an appealing mechanism for development of the DAF eskers because, given geomorphologic constraints on the minimum thickness of the DAF ice sheet and evidence for higher geothermal heat fluxes at the Noachian-Hesperian boundary (Montési and Zuber, 2003; Solomon et al., 2005), prior work (Fastook et al., 2012) suggests that it does not require a substantially warmer climate. However, new UMISM simulations, which account for the shape and extent of the south polar ice sheet in an early Mars with a thick CO2 atmosphere, suggest that basal melting of the DAF ice sheet is negligible except in substantially warmer climates. Furthermore, modeled basal temperatures are highest near the center of the ice sheet, where ice is thickest, rather than in the distal regions of the DAF where eskers are observed. Generation of the DAF eskers by volcano-ice interaction (Figure 31d) shares the advantage of not requiring a substantially warmer climate for early Mars, and the large meltwater fluxes indicated by esker morphometry in the DAF are frequently observed in Icelandic glaciovolcanic jökulhlaups (e.g. Tómasson, 1996, 2002; Maizels, 1997; Baker, 2009). The Dorsa Argentea esker population (Figure 1) is located near (though not directly adjoining) both a population of irregular craters that could plausibly have 85 originated from explosive glaciovolcanism (Figure 32) and the Australe Montes, which lack obvious vents but may also be volcanic in origin; they are collinear with both the volcanic Sisyphi Montes and the possible vents, and their relationship to surrounding topography is inconsistent with an origin as erosional remnants of a crater rim. Eskers (specifically, the Dorsa Argentea population) emerge from only one downslope aspect of the Australe Mons and the possible explosive vent structures, however, and extend hundreds of kilometers away from them. The Dorsa Argentea eskers are also oriented north-south in their southernmost reaches, which suggests that the meltwater that flowed through these channels in their source regions originated south of the Dorsa Argentea, rather than to the east near Australe Montes, Sisyphi Montes, or the possible vent structures. Furthermore, very few eskers are present near the glaciovolcanic Sisyphi Montes (Figure 1), and no plausible glaciovolcanic structures are associated with the eskers near 90°W or 270°W in the DAF. Elevated geothermal heat due to subglacial intrusive volcanism cannot be ruled out, but would likely have been at a maximum in the same regions as the extrusive glaciovolcanic edifices, and no positive evidence for intrusive volcanism is present in our datasets. Additional evidence for subglacial volcanism (and additional eskers) could be buried beneath Amazonian polar deposits, but the Dorsa Argentea themselves would still be unaccounted for. The remaining possibility is that the DAF eskers, like most eskers on Earth, were constructed in segments beneath the wet-based margins of an ice sheet undergoing stagnant retreat. The large meltwater fluxes and short construction times inferred for the DAF eskers are both consistent with this formation mechanism. Furthermore, of all the mechanisms considered, wet-based retreat is best able to explain the locations of the 86 eskers in the regions of the DAF where ice is first removed. Difficulties with this mechanism include (1) the preservation and continuity of the DAF eskers, (2) the paucity of dendritic channels radiating from the edges of the DAF where proglacial sandur might be expected in this scenario, if substantial top-down melting occurred, and (3) raising martian global temperatures to a point where substantial melting can occur at the poles. Each of these potential problems can be addressed, however. With regard to preservation and continuity, the eskers spanning much of Canada, which are believed to have formed sequentially behind retreating margins of the Laurentide Ice Sheet, are continuous over similar lengths (e.g. Storrar et al., 2014a; 2014b), and are well-preserved in regions where ice rarely readvanced during deglaciation (Storrar et al., 2014a). If retreat of the DAF ice sheet occurred during long-term cooling or drying of Mars at the Noachian-Hesperian transition, few readvances would be expected, as UMISM results show that the extent of the DAF ice sheet in warm climates is particularly sensitive to ice inventory. With regard to marginal subaerial channels, first, our results indicate that substantial basal melting at the edges of the ice sheet can occur in a climate that is not warm enough for top-down melting to occur except for a few days per year in small and scattered regions of the ice sheet; numerous supraglacial channels may therefore never have formed. Furthermore, even if the eskers were generated in a warmer climate where top-down melting was more common, few channels are observed at the boundaries of wet-based ice sheets undergoing top-down melting. Instead, supraglacial channels in wet- based regions of ice sheets often terminate on the glacier surface in moulins or crevasses, with all further downstream drainage occuring through the subglacial drainage network 87 (e.g. Hammer and Smith, 1983; Marston, 1983; Greenwood et al., 2007; Syverson and Mickelson, 2009; Bartholomew et al., 2011). Any supraglacial channels on the DAF ice sheet that terminated before the glacier margin would not have been preserved when the ice into which they were carved eventually ablated completely. The five large channels emerging from the edges of the DAF (Ghatan et al., 2004), the channels connecting the breached craters along the eastern edge of the 0°W lobe of the DAF (Milkovich et al., 2002), and the channel emerging from the Argentea Planum paleolake (Dickson and Head, 2006) have no visible tributaries. We propose that these channels were the main branches of dendritic valley networks heading either on the glacier surface or in non- erosive subglacial drainage systems. Finally, while a geologically plausible mechanism for warming early Mars to the point where temperatures rise above freezing at the poles has yet to be conclusively determined, we have estimated from geomorphology that the largest eskers in the DAF would have taken between ~150 days and ~320 years to form. Our GCM simulations show that DAF ice melts back past the esker-forming regions in ~200 years in a climate where surface temperatures exceed the 1000 mb pure CO2 baseline case by 40-45° in the south polar regions. It has been proposed that the equatorial valley networks formed during brief warm periods in an otherwise cold climate, where water migrated as ice to cold traps in the martian highlands between periods of warming (Wordsworth et al., 2013). We propose that the DAF eskers formed during the same episodes of warming. The crater exposure ages for the esker populations (Kress and Head, 2015), which fall within the range of crater ages for the valley networks (Fassett and Head, 2008), are consistent with this scenario. 88 Milkovich et al. (2002) calculated that at least 1012 m3 of water ponded in the open-basin lakes east of the 0°W lobe of the DAF. If the channels leading into these lakes were fed at comparable rates to the subglacial channels that left the largest DAF eskers (i.e. with discharges on the order of 103 – 105 m3 s-1), this implies that at least one warm interval lasted between hundreds of days and tens of years. We note that surface temperatures are above freezing for most of the year at the poles in our warm endmember simulations, but would only need to exceed freezing seasonally for top-down melting to occur. The available data do not allow us to determine a preferred model for the later, cold-based retreat of the DAF ice sheet. Our results (Section 3.2) suggest that ~50 ky of very high-obliquity excursions or tens of millions of years at an atmospheric pressure of 200 mb would be sufficient to remove ~1800 m of south polar ice. Crater counts on DAF esker populations suggest that ~300 million years passed between the removal of the ice in Promethei Planum and ice in the vicinity of the Dorsa Argentea. Spin-axis obliquity > 60° has been rare over martian history (Laskar et al., 2004), but need not have been frequent to account for removal of the DAF by this mechanism; 50 ky is less than two hundredths of one percent of 300 My. Atmospheric pressure must have been lost over time for the DAF to have been stable at the Noachian-Hesperian boundary and not in the present day, and the gradual loss would have resulted in slower (hence more consistent with observations) removal rates than in our simulations. Glacial flow modeling results indicate that the extent of the DAF ice sheet also decreases with global ice inventory. All three mechanisms (obliquity variations, atmospheric loss, and depletion of the global water inventory) are likely to have contributed to the ablation of the DAF ice sheet. 89 In glacial flow model simulations, a south circumpolar ice sheet with a wavenumber-3 planform occurs with a 1000 mb CO2 atmosphere. The maximum thickness of the modeled ice sheet is generally consistent with geomorphologic observations for all ice inventories studied; maximum thicknesses are 1500 – 2000 m regardless of climate, geothermal heat flux, or ice inventory, and the Sisyphi Montes appear to have formed in an ice sheet at least ~1800 m thick (Ghatan and Head, 2002). The extent of the modeled ice sheet is most similar to observations in our simulations with a global ice inventory of ~137 m GEL and an idealized greenhouse gas providing ~38 – 48° of greenhouse warming at the poles (in addition to the warming provided by 1 bar of CO2). Contrary to earlier findings, which used ice accumulation and surface temperature distributions more similar to those of modern Mars (Fastook et al., 2012), basal melting does not occur under any of the climate and geothermal heat flux scenarios studied other than the simulations with additional greenhouse warming. A salient feature in south polar MARSIS (Mars Advanced Radar for Subsurface and Ionospheric Sounding) radar data is a subsurface reflector whose shape and extent correlate closely with those of the DAF (Plaut et al., 2007). This reflector lies 600 – 900 m beneath the surface, and Plaut et al. (2007) interpret it to represent an interface between ice-poor bedrock and an ice-rich overlying regolith layer. This layer could represent debris-rich remnant ice from the DAF ice sheet, or, potentially, regions where basal meltwater from the DAF ice sheet infiltrated the underlying ground and is preserved as pore ice. In light of the distinct, lobate morphology of the plains material, the occurrence of isolated plains material in craters just outside the boundary of the DAF, and the similarity between the depth of the MARSIS reflector and the apparent depth of 90 the DAF plains material, we favor the interpretation that the DAF plains material formed from downwasting of the DAF ice sheet and still contains some volatiles. 6. Conclusions We have presented geomorphometric evidence that flux through the eskers in the Dorsa Argentea Formation was on the order of 102 – 104 m3 s-1. Minimum formation times estimated from esker dimensions are on the order of hundreds of days to hundreds of years. In GCM simulations with a thick (600 or 1000 mb) pure CO2 atmosphere, an extensive region of ice accumulation forms near the south pole, with DAF-like lobes along the 0°W and 90°W meridians. The shape of the ice sheet is due to the regional topography of the southern polar regions, and the dependence of surface temperature upon surface elevation in a thicker atmosphere. This indicates that (1) the asymmetry of the DAF is therefore likely to reflect an asymmetric ice sheet rather than asymmetric erosion of the glacial deposits and (2) elevated topography along the 90°W meridian, i.e. the Tharsis bulge, was required at the time the DAF formed in order to generate an ice sheet with a lobe extending in that direction. In glacial flow model simulations, basal melting does not occur in any of the pure-CO2 atmospheres studied, regardless of the value used for the the late Noachian – early Hesperian geothermal heat flux, but does occur in simulations in which an idealized greenhouse gas provides additional warming beyond that from CO2 alone. Furthermore, the locations of maximum modeled basal melting are similar to the locations of observed glaciofluvial landforms in simulations with additional greenhouse warming, but are close 91 to the center of the ice sheet in pure-CO2 simulations. The extent of the ice sheet is also much smaller than that of the DAF in the pure-CO2 simulations, but closely matches that of the DAF in simulations with an additional 38 – 48° of greenhouse warming in the vicinity of the DAF. We therefore conclude that additional greenhouse warming beyond that from CO2 alone was required to form the eskers in the DAF and the extensive plains that comprise the present-day DAF; we hypothesize that basal melting of the DAF ice sheet occurred during the same warm intervals on early Mars that allowed the formation of the equatorial valley networks. Crater ages for the esker populations (Kress and Head, 2015) are consistent with this hypothesis. Acknowledgements We thank the NASA Graduate Student Researchers Program (Grant NNX12AI39H) for financial support to KES, Robin Wordsworth for substantial and ongoing input to this chapter, and Anne Coté for assistance with Figure 31. References Ackiss, S. E., and J. Wray (2014), Occurrences of possible hydrated sulfates in the southern high latitudes of Mars, Icarus, 243, 311-324. Allen, C. C. (1979), Volcano-ice interactions on Mars, Journal of Geophysical Research: Solid Earth (1978–2012), 84(B14), 8048-8059. Anderson, S. P., S. A. Longacre, and E. R. Kraal (2003), Patterns of water chemistry and discharge in the glacier-fed Kennicott River, Alaska: evidence for subglacial water storage cycles, Chemical Geology, 202, 297–312. 92 Baker, V. R. (2009), High-energy megafloods: planetary settings and sedimentary dynamics. Flood and megaflood processes and deposits: recent and ancient examples, 32, 3-15. Baker, V. R., R. G. Strom, V. C. Gulick, J. S. Kargel, G. Komatsu, and V. S. Kale (1991), Ancient Oceans, Ice Sheets and the Hydrological Cycle on Mars, Nature, 352(6336), 589-594. Banks, M. E., N. P. Lang, J. S. Kargel, A. S. McEwen, V. R. Baker, J. A. Grant, J. D. Pelletier, and R. G. Strom (2009), An analysis of sinuous ridges in the southern Argyre Planitia, Mars using HiRISE and CTX images and MOLA data, J Geophys Res-Planet, 114. Barnhart, C. J., A. D. Howard, and J. M. Moore (2009), Long-term precipitation and late- stage valley network formation: Landform simulations of Parana Basin, Mars, Journal of Geophysical Research: Planets, 114, E01003. Bartholomaus, T. C., R. S. Anderson, and S. P. Anderson (2011), Growth and collapse of the distributed subglacial hydrologic system of Kennicott Glacier, Alaska, USA, and its effects on basal motion, Journal of Glaciology, 57, 985–1002. Bartholomew, I., P. Nienow, A. Sole, D. Mair, T. Cowton, S. Palmer, and J. Wadham (2011). Supraglacial forcing of subglacial drainage in the ablation zone of the Greenland ice sheet, Geophysical Research Letters 38, 8. Bernhardt, H., H. Hiesinger, D. Reiss, M. Ivanov, and G. Erkeling (2013), Putative eskers and new insights into glacio-fluvial depositional settings in southern Argyre Planitia, Mars, Planet Space Sci, 85, 261-278. 93 Bingham, R. G., Nienow, P. W., Sharp, M. J., & Boon, S. (2005). Subglacial drainage processes at a High Arctic polythermal valley glacier. Journal of Glaciology, 51(172), 15-24. Bouley, S., Baratoux, D., Matsuyama, I., Forget, F., Costard, F., & Séjourné, A. (2015). True Polar Wander Recorded by the Distribution of Martian Valley Networks. Lunar and Planetary Science Conference 46, abstract #1887. Brennand, T. A. (1994), Macroforms, Large Bedforms and Rhythmic Sedimentary Sequences in Subglacial Eskers, South-Central Ontario - Implications for Esker Genesis and Meltwater Regime, Sediment Geol, 91(1-4), 9-55. Brennand, T. A. (2000), Deglacial meltwater drainage and glaciodynamics: inferences from Laurentide eskers, Canada, Geomorphology, 32(3-4), 263-293. Burke, M. J., T. A. Brennand, and A. J. Perkins (2012), Transient subglacial hydrology of a thin ice sheet: insights from the Chasm esker, British Columbia, Canada, Quaternary Science Reviews 58, 30-55. Burke, M. J., T. A. Brennand, and D. B. Sjogren (2015), The role of sediment supply in esker formation and ice tunnel evolution, Quaternary Science Reviews 115, 50- 77. Carr, M. H., and J. W. Head III (2015), Martian surface/near-surface water inventory: Sources, sinks, and changes with time, Geophys. Res. Lett., 42, 1-7, doi: 10.1002/2014GL062464. Clark, P. U., and J. S. Walder (1994), Subglacial drainage, eskers, and deforming beds beneath the Laurentide and Eurasian ice sheets, Geological Society of America Bulletin 106, 2, 304-314. 94 Colaprete, A., J. R. Barnes, R. M. Haberle, J. L. Hollingsworth, H. H. Kieffer, and T. N. Titus (2005), Albedo of the south pole on Mars determined by topographic forcing of atmosphere dynamics, Nature 435, 7039, 184-188. Condit, C. D., L. A. Soderblom (1978), Geologic map of the Mare Australe area of Mars, U.S. Geol. Surv. Misc. Invest. Ser. Map, I-1706-C. Cummings, D. I., G. Gorrell, J.-P. Guilbault, J. A. Hunter, C. Logan, D. Ponomarenko, J.- M. P. André, S. E. Pullan, H. A. J. Russell, and D. R. Sharpe (2011), Sequence stratigraphy of a glaciated basin fill, with a focus on esker sedimentation, Geological Society of America Bulletin, 123, 1478-1496. Dickson, J., and J. W. Head (2006), Evidence for an Hesperian-aged South Circum-Polar Lake Margin Environment on Mars, Planet Space Sci, 54(3), 251-272. Ensminger, S. L., E. B. Evenson, R. B. Alley, G. J. Larson, D. E. Lawson, and J. C. Strasser (1999). Example of the dependence of ice motion on subglacial drainage system evolution: Matanuska Glacier, Alaska, United States, Geological Society of America Special Papers 337, 11-22. Fassett, C. I., and J. W. Head (2008), The timing of martian valley network activity: Constraints from buffered crater counting, Icarus, 195, 61-89, doi:10.1016/j.icarus.2007.12.009. Fastook, J. L., J. W. Head, D. R. Marchant, and F. Forget (2008), Tropical mountain glaciers on Mars: Altitude-dependence of ice accumulation, accumulation conditions, formation times, glacier dynamics, and implications for planetary spin-axis/orbital history, Icarus, 198, 305-317, doi:10.1016/j.icarus.2008.08.008. 95 Fastook, J. L., J. W. Head III, D. R. Marchant, F. Forget, and J.-B. Madeleine (2012), Early Mars climate near the Noachian-Hesperian boundary: Independent evidence for cold conditions from basal melting of the south polar ice sheet (Dorsa Argentea Formation) and implications for valley network formation, Icarus, 219, 25-40. Fastook, J. L., and J. W. Head III (2015), Glaciation in the Late Noachian Icy Highlands: Ice accumulation, distribution, flow rates, basal melting, and top-down melting rates and patterns, Planet. Space Sci., 106, 82-98, doi: 10.1016/j.pss.2014.11.028. Flowers, G. E., H. Björnsson, and F. Pálsson (2003). New insights into the subglacial and periglacial hydrology of Vatnajökull, Iceland, from a distributed physical model, Journal of Glaciology 49, 165, 257-270. Forget, F., R. M. Haberle, F. Montmessin, B. Levrard, J. W. Head (2006), Formation of Glaciers on Mars by Atmospheric Precipitation at High Obliquity, Science, 311, 368-371. Forget, F., R. Wordsworth, E. Millour, J. B. Madeleine, L. Kerber, J. Leconte, E. Marcq, and R. M. Haberle (2013), 3D modelling of the early martian climate under a denser CO2 atmosphere: Temperatures and CO2 ice clouds, Icarus, 222(1), 81- 99. Ghatan, G. J., and J. W. Head (2002), Candidate subglacial volcanoes in the south polar region of Mars: Morphology, morphometry, and eruption conditions, J Geophys Res-Planet, 107(E7). 96 Ghatan, G. J., and J. W. Head (2004), Regional drainage of meltwater beneath a Hesperian-aged south circumpolar ice sheet on Mars, J Geophys Res-Planet, 109(E7). Ghatan, G. J., J. W. Head, and S. Pratt (2003), Cavi Angusti, Mars: Characterization and assessment of possible formation mechanisms, J Geophys Res-Planet, 108(E5). Ghatan, G. J., J. W. Head, and L. Wilson (2005), Mangala Valles, Mars: Assessment of early stages of flooding and downstream flood evolution, Earth Moon Planets, 96(1-2), 1-57. Gough, D. O. (1981). Solar interior structure and luminosity variations. In: Physics of Solar Variations (pp. 21-34). Springer Netherlands. Greenwood, S. L., C. D. Clark, and A. L. C. Hughes (2007), Formalising an inversion methodology for reconstructing ice sheet retreat patterns from meltwater channels: application to the British Ice Sheet, Journal of Quaternary Science 22, 6, 637-645. Hammer, K. M., and N. D. Smith (1983), Sediment production and transport in a proglacial stream: Hilda Glacier, Alberta, Canada, Boreas 12, 2, 91-106. Head, J. W., and S. Pratt (2001), Extensive Hesperian-aged south polar ice sheet on Mars: Evidence for massive melting and retreat, and lateral flow and pending of meltwater, J Geophys Res-Planet, 106(E6), 12275-12299. Head, J. W., and B. Hallet (2001a), Origin of Sinuous Ridges in the Dorsa Argentea Formation: New Observations and Tests of the Esker Hypothesis, in Lunar and Planetary Science Conference, edited, p. 1373. 97 Head, J. W., and B. Hallet (2001b), Origin of Sinuous Ridges in the Dorsa Argentea Formation: Additional Criteria for Tests of the Esker Hypothesis, in Lunar and Planetary Science Conference, edited, p. 1366. Head, J. W., and D. R. Marchant (2003), Cold-based mountain glaciers on Mars: Western Arsia Mons, Geology, 31, 641-644. Hoke, M. R. T., B. M. Hynek, and G. E. Tucker (2011), Formation timescales of large Martian valley networks, Earth and Planetary Science Letters, 312, 1–12. Hooke, R. L., and J. Fastook (2007), Thermal conditions at the bed of the Laurentide ice sheet in Maine during deglaciation: implications for esker formation, Journal of Glaciology, 53, 646-658. Howard, A. D. (1981), Etched plains and braided ridges of the south polar region of Mars: features produced by basal melting of ground ice?Rep., 286-288 pp. Irwin, R. P., R. A. Craddock, and A. D. Howard (2005), Interior channels in Martian valley networks: Discharge and runoff production, Geology, 33 (6), 489–492. Kargel, J. S., and R. G. Strom (1992), Ancient Glaciation on Mars, Geology, 20(1), 3-7. Kargel, J. S., and R. G. Strom (1996), Global climatic change on Mars - Today a frozen world, Mars at one time may have had more temperate conditions; with flowing rivers, thawing seas, melting glaciers and, perhaps, abundant life, Sci Am, 275(5), 80-88. Kerber, L., Forget, F., Madeleine, J. B., Wordsworth, R., Head, J. W., & Wilson, L. (2013). The effect of atmospheric pressure on the dispersal of pyroclasts from Martian volcanoes. Icarus, 223(1), 149-156. 98 Kerber, L., Forget, F., and Wordsworth, R. (2015), Sulfur in the early martian atmosphere revisited: Experiments with a 3-D Global Climate Model, Icarus, 261, 133–148. Kieffer, H. H. (2013), Thermal model for analysis of Mars infrared mapping, Journal of Geophysical Research: Planets, 118, 451-470. Kite, E. S., J.-P. Williams, A. Lucas, and O. Aharonson (2014), Paleopressure of Mars’ atmosphere from small ancient craters, Nature Geoscience, 7, 335–339. Kite, E. S., I. Matsuyama, M. Manga, J. T. Perron, and J. X. Mitrovica (2009), True Polar Wander driven by late-stage volcanism and the distribution of paleopolar deposits on Mars, Earth Planet Sc Lett, 280(1-4), 254-267. Kleman, J. (1994), Preservation of Landforms under-Ice Sheets and Ice Caps, Geomorphology, 9(1), 19-32. Komar, P. D. (1979). Comparisons of the hydraulics of water flows in Martian outflow channels with flows of similar scale on Earth. Icarus, 37(1), 156-181. Kress, A. M., and J. W. Head (2015), Late Noachian and early Hesperian ridge systems in the south circumpolar Dorsa Argentea Formation, Mars: Evidence for two stages of melting of an extensive late Noachian ice sheet, Planet Space Sci, 109, 1-20. Lammer, H., E. Chassefière, Ö. Karatekin, A. Morschhauser, P. B. Niles, O. Mousis, P. Odert, U. V. Möstl, D. Breuer, V. Dehant, M. Grott, H. Gröller, E. Hauber, and L. B. S. Pham (2013), Outgassing History and Escape of the Martian Atmosphere and Water Inventory, Space Science Reviews, 174, 113–154. 99 Lappegard, G., J. Kohler, M. Jackson, and J. O. Hagen (2006). Characteristics of subglacial drainage systems deduced from load-cell measurements, Journal of Glaciology 52, 176, 137-148. Laskar, J., A. Correia, M. Gastineau, F. Joutel, B. Levrard, and P. Robutel (2004), Long term evolution and chaotic diffusion of the insolation quantities of Mars, Icarus, 170(2), 343-364. Leopold, L. B., Wolman, M. G., & Miller, J. P. (1964). Fluvial processes in geomorphology The University of Chicago Press San Francisco. Lorang, M. S., & Hauer, F. R. (2003). Flow competence and streambed stability: an evaluation of technique and application. Journal of the North American Benthological Society, 22(4), 475-491. Madeleine, J. B., F. Forget, J. W. Head, B. Levrard, F. Montmessin, and E. Millour (2009), Amazonian northern mid-latitude glaciation on Mars: A proposed climate scenario, Icarus, 203, 390-405. Maizels, J. (1997), Jökulhlaup deposits in proglacial areas, Quaternary Science Reviews, 16, 793-819. Marston, R. A. (1983), Supraglacial stream dynamics on the Juneau Icefield, Annals of the Association of American Geographers 73, 4, 597-608. Milkovich, S. M., J. W. Head, and S. Pratt (2002), Meltback of Hesperian-aged ice-rich deposits near the south pole of Mars: Evidence for drainage channels and lakes, J Geophys Res-Planet, 107(E6). 100 Mischna, M. A., V. Baker, R. Milliken, M. Richardson, and C. Lee (2013), Effects of obliquity and water vapor/trace gas greenhouses in the early martian climate, J Geophys Res-Planet, 118(3), 560-576. Montési, L. G. J., and M. T. Zuber (2003). Clues to the lithospheric structure of Mars from wrinkle ridge sets and localization instability, Journal of Geophysical Research: Planets (1991–2012) 108, E6. Mooers, H. D. (1990), A glacial-process model: The role of spatial and temporal variations in glacier thermal regime, Geological Society of America Bulletin 102, 2, 243-251. Murchie, S. L., et al. (2009), A synthesis of Martian aqueous mineralogy after 1 Mars year of observations from the Mars Reconnaissance Orbiter, J Geophys Res- Planet, 114. Murray, B. C., L. A. Soderblom, J. A. Cutts, R. P. Sharp, D. J. Milton, and R. B. Leighton (1972), Geological framework of the south polar region of Mars, Icarus, 17, 328–345. Peterson, J. E. (1977), Geologic map of the Noachis quadrangle of Mars, U.S. Geol. Surv. Misc. Invest. Ser. Map, I-910. Plaut, J. J., A. Ivanov, A. Safaeinili, S. M. Milkovich, G. Picardi, R. Seu, and R. Phillips (2007), Radar sounding of subsurface layers in the south polar plains of Mars: correlation with the Dorsa Argentea Formation, In Lunar and Planetary Science Conference, vol. 38, p. 2144. Read, P. L., and S. R. Lewis (2004), The martian climate revisited: Atmosphere and environment of a desert planet, Springer Science & Business Media. 101 Scott, D. H., and M. H. Carr (1978), Geologic map of Mars, U.S. Geol. Surv. Misc. Invest. Ser. Map, I-1083. Skinner, J. A., J. M. Dohm, R. P. Irwin III, E. J. Kolb, C. M. Fortezzo, T. Platz, G. G. Michael, and T. Hare (2014). Geologic map of Mars. US Department of the Interior, US Geological Survey. Smith, D., G. Neumann, R. E. Arvidson, E. A. Guinness, and S. Slavney (2003), "Mars Global Surveyor Laser Altimeter Mission Experiment Gridded Data Record", NASA Planetary Data System, MGS-M-MOLA-5-MEGDR-L3-V1.0. Solomon, S.C. et al., 2005. New perspectives on ancient Mars. Science 307, 1214–1220. Soto, A., M. Mischna, T. Schneider, C. Lee, and M. Richardson (2015), Martian atmospheric collapse: Idealized GCM studies, Icarus 250, 553-569. Storrar, R. D., C. R. Stokes, and D. J. A. Evans (2014a), Morphometry and pattern of a large sample (> 20,000) of Canadian eskers and implications for subglacial drainage beneath ice sheets, Quaternary Science Reviews 105, 1-25. Storrar, R. D., C. R. Stokes, and D. J. A. Evans (2014b), Increased channelization of subglacial drainage during deglaciation of the Laurentide Ice Sheet, Geology, 42, 239-242. Syverson, K. M., and D. M. Mickelson (2009), Origin and significance of lateral meltwater channels formed along a temperate glacier margin, Glacier Bay, Alaska, Boreas 38, 1, 132-145. Tanaka, K. L., D. H. Scott, Geological Survey (U.S.), United States. National Aeronautics and Space Administration., and Planetary Geology and Geophysics 102 Program (U.S.) (1987), Geologic map of the polar regions of Mars, Geological Survey (U.S.), Reston, Va. Tanaka, K. L., & Kolb, E. J. (2001). Geologic history of the polar regions of Mars based on Mars Global Surveyor Data: I. Noachian and Hesperian Periods. Icarus, 154(1), 3-21. Tanaka, K. L., Robbins, S. J., Fortezzo, C. M., Skinner, J. A., & Hare, T. M. (2014). The digital global geologic map of Mars: Chronostratigraphic ages, topographic and crater morphologic characteristics, and updated resurfacing history. Planetary and Space Science, 95, 11-24. Tómasson, H. (1996), The jökulhlaup from Katla in 1918, Annals of Glaciology, 22, 249- 254. Tómasson, H. (2002), Catastrophic floods in Iceland, International Association of Hydrological Sciences Publication no. 271, 121-128. Urata, R. A., and O. B. Toon (2013), Simulations of the martian hydrologic cycle with a general circulation model: Implications for the ancient martian climate, Icarus, 226(1), 229-250. Wadham, J. L., R. Hodgkins, R. J. Cooper, and M. Tranter (2001). Evidence for seasonal subglacial outburst events at a polythermal glacier, Finsterwalderbreen, Svalbard, Hydrological processes 15, 12, 2259-2280. Wordsworth, R., F. Forget, E. Millour, J. W. Head, J. B. Madeleine, and B. Charnay (2013), Global modelling of the early martian climate under a denser CO2 atmosphere: Water cycle and ice evolution, Icarus, 222(1), 1-19. 103 Wordsworth, R. D., Kerber, L., Pierrehumbert, R. T., Forget, F., & Head, J. W. (2015). Comparison of “warm and wet” and “cold and icy” scenarios for early Mars in a 3-D climate model. Journal of Geophysical Research: Planets, 120(6), 1201-1219. Wray, J. J., S. L. Murchie, S. W. Squyres, F. P. Seelos, and L. L. Tornabene (2009), Diverse aqueous environments on ancient Mars revealed in the southern highlands, Geology, 37(11), 1043-1046. 104 Table 1. Morphometry and estimated formative discharges for the Dorsa Argentea and Parva Planum esker populations. Dorsa Argentea Parva Planum Width 2000 m 1500 m Slope 1.331 m km-1 1.750 m km-1 Q(1m, sand) 1600 m3 s-1 1400 m3 s-1 Q(10m, sand) 65000 m3 s-1 56000 m3 s-1 Q(1m, gravel) 650 m3 s-1 560 m3 s-1 Q(10m, gravel) 35000 m3 s-1 30000 m3 s-1 105 Table 2. Time to construct Dorsa Argentea Formation eskers for varying estimates of ice debris content and formative discharge. Dorsa Argentea Debris Q 6.5 × 102 m3s-1 6.5 × 103 m3s-1 6.5 × 104 m3s-1 1% ~320 years ~32 years ~3.2 years 10% ~32 years ~3.2 years ~114 days Parva Planum Debris Q 5.6 × 102 m3s-1 5.6 × 103 m3s-1 5.6 × 104 m3s-1 1% ~210 years ~21 years ~2.1 years 10% ~21 years ~2.1 years ~75 days 106 Figure 1. Overview of the Dorsa Argentea Formation. Geologic units are as mapped by Tanaka and Scott (1987): the present-day south polar cap is shown in white, the Amazonian polar deposits in blue, and the DAF in orange and gold. Cavi are shown in black, eskers in dark blue, large fluvial channels in light blue, breached crater lakes in green, the Argentea Planum paleolake in purple, Sisyphi Montes in red, Australe Montes in pink, and possible explosive glaciovolcanic vents in dark red. 107 Figure 2. Schematic representation of possible trajectories for the development of eskers beneath, and the subsequent ablation of, a large south circumpolar ice sheet at the Noachian-Hesperian boundary. 108 Figure 3. Annual average surface temperature (°K) in the southern hemisphere with spin- axis obliquity of 25° and global CO2 surface pressure of (a) 600 mb, (b) 1000 mb, and (c) 1500 mb. a b 109 c 110 Figure 4. Annual average surface temperature (°K) in the southern hemisphere with spin- axis obliquity of 41.8° and global CO2 surface pressure of (a) 600 mb, (b) 1000 mb, and (c) 1500 mb. a b 111 c 112 Figure 5. Annual average surface temperature (°K) in the region of the Dorsa Argentea Formation (90°S to 55°S) with spin-axis obliquity of 25° and (a) the Tharsis bulge flattened, (b) Hellas and Argyre basins removed, and (c) globally flat topography. a b 113 c 114 Figure 6. Scatter plots of annual average surface temperature as a function of altitude for each grid point in GCM simulations with 25° (a) and 41.8° (b) spin-axis obliquity. Red points are from runs with 600 mb CO2 atmospheres, green points are from runs with 1000 mb CO2 atmospheres, and blue points are from runs with 1500 mb CO2 atmospheres. a b 115 Figure 7. Annual average surface temperature (°K) contours (white lines) superimposed on geologic map showing the Dorsa Argentea eskers (dark blue lines) in a 1000 mb CO2 atmosphere at 25° spin-axis obliquity (a) and 41.8° spin-axis obliquity (b). a b 116 Figure 8. Annual maximum surface temperatures (°K) in a GCM simulation with a 1500 mb CO2 atmosphere and 41.8° spin-axis obliquity. 117 Figure 9. Ice accumulation (positive values) and ablation (negative values) after ten model years (~400 years of ice evolution) in the region of the DAF (90°S to 55°S) for 25° spin-axis obliquity and average CO2 surface pressure of (a) 600 mb, (b) 1000 mb and (c) 1500 mb. a b 118 c 119 Figure 10. Ice accumulation (positive values) and ablation (negative values) after ten model years (~400 years of ice evolution) in the region of the DAF (90°S to 55°S) for 41.8° spin-axis obliquity and average CO2 surface pressure of (a) 600 mb, (b) 1000 mb and (c) 1500 mb. a b 120 c 121 Figure 11. Ice accumulation (positive values) and ablation (negative values) after ten model years (~400 years of ice evolution) in the region of the DAF (90°S to 55°S) under increased solar flux (1761.9 W m-2 at 1 AU) and average CO2 surface pressure of 1000 mb at (a) 25° spin-axis obliquity and (b) 41.8° spin-axis obliquity. a b 122 Figure 12. Ice accumulation (positive values) and ablation (negative values) of a flat ice sheet with the spatial extent of the DAF after ten model years (~400 years of ice evolution) for (a) 60° spin-axis obliquity and a 1000 mb CO2 atmosphere, (b) 70° spin- axis obliquity and a 1000 mb CO2 atmosphere, (c) 60° spin-axis obliquity and a 200 mb CO2 atmosphere, and (d) 70° spin-axis obliquity and a 200 mb CO2 atmosphere. a b 123 c d 124 Figure 13. Ice accumulation (positive values) and ablation (negative values) of a flat ice sheet with the spatial extent of the DAF after ten model years (~400 years of ice evolution) for atmospheric pressures of 200 mb (a) and 50 mb (b). a b 125 Figure 14. Difference in surface water ice (in kg m-2) after ten model years (~400 years of ice evolution) in a 50 mb CO2 atmosphere with ice albedo set equal to ground albedo, showing net ablation over most of the DAF region. 126 Figure 15. Starting from an ice sheet with the areal extent of the DAF, location of the ~20 m ice contour after ~100 years (purple contour), ~200 years (red contour), ~300 years (orange contour), and ~400 years (yellow contour) of ice ablation under increased solar flux (1761.9 W m-2 at 1 AU), 25° spin-axis obliquity, and average CO2 surface pressure of 1000 mb. Contours are overlaid on a map of MOLA 512 pixel per degree surface topography (shaded) and DAF eskers (blue lines). 127 Figure 16. Equilibrium surface topography (in m) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, ice inventory of ~275 m GEL, and climate data from GCM simulations with (a) a 600 mb CO2 atmosphere and (b) a 1000 mb CO2 atmosphere. a b 128 Figure 17. Equilibrium surface topography (in m) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, climate data from a GCM run with a 1000 mb atmosphere, and ice inventories (a) ~137 m GEL, (b) ~275 m GEL, (c) ~550 m GEL, and (d) ~1375 m GEL. a b 129 c d 130 Figure 18. Equilibrium surface topography (in m) of the DAF ice sheet in simulations with an ice inventory of ~137 m GEL, climate data from a GCM run with a 1000 mb CO2 atmosphere, and geothermal heat flux of (a) 45 mW m-2 and (b) 65 mW m-2. a 131 Figure 19. Flow velocity (in log10 mm yr-1) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, ice inventory of ~275 m GEL, and climate data from GCM simulations with (a) a 600 mb CO2 atmosphere and (b) a 1000 mb CO2 atmosphere. a b 132 Figure 20. Basal melting rate (in mm yr-1) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, ice inventory of ~275 m GEL, and climate data from GCM simulations with (a) a 600 mb CO2 atmosphere and (b) a 1000 mb CO2 atmosphere. a b 133 Figure 21. Basal melting rate (in mm yr-1) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, climate data from a GCM run with a 1000 mb atmosphere, and ice inventories (a) ~137 m GEL, (b) ~275 m GEL, (c) ~550 m GEL, and (d) ~1375 m GEL. a b 134 c d 135 Figure 22. Basal melting rate (in mm yr-1) of the DAF ice sheet in simulations with an ice inventory of ~137 m GEL, climate data from a GCM run with a 1000 mb CO2 atmosphere, and geothermal heat flux of (a) 45 mW m-2 and (b) 65 mW m-2. a b 136 Figure 23. Basal temperature (in °K) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, ice inventory of ~275 m GEL, and climate data from GCM simulations with (a) a 600 mb CO2 atmosphere and (b) a 1000 mb CO2 atmosphere. a b 137 Figure 24. Basal temperature (in °K) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, climate data from a GCM run with a 1000 mb atmosphere, and ice inventories (a) ~137 m GEL, (b) ~275 m GEL, (c) ~550 m GEL, and (d) ~1375 m GEL. a b 138 c d 139 Figure 25. Difference (°K) between annual average surface temperatures in the GCM simulations with added “grey gas” absorption and the baseline 1000 mb CO2 run, shown for grey gas κ = 2.5 × 10-5 (a), 5.0 × 10-5 (b), 1.0 × 10-4 (c), 1.5 × 10-4 (d), and 2.0 × 10-4 (e). a b 140 c d 141 e 142 Figure 26. Equilibrium surface topography (in m) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, ice inventory of ~275 m GEL, and climate data from GCM simulations with a 1000 mb CO2 atmosphere and grey gas κ = 2.5 × 10-5 (a), 5.0 × 10-5 (b), 1.0 × 10-4 (c), 1.5 × 10-4 (d), and 2.0 × 10-4 (e). a 143 b 144 c 145 d 146 e 147 Figure 27. Equilibrium surface topography (in m) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, climate data from a GCM run with a 1000 mb atmosphere and grey gas κ = 1.0 × 10-4, and ice inventories (a) ~137 m GEL, (b) ~275 m GEL, and (c) ~550 m GEL. a 148 b 149 c 150 Figure 28. Basal melting rate (in mm yr-1) of the DAF ice sheet in simulations with geothermal heat flux of 55 mW m-2, ice inventory of ~275 m GEL, and climate data from GCM simulations with a 1000 mb CO2 atmosphere and grey gas κ = 2.5 × 10-5 (a), 5.0 × 10-5 (b), 1.0 × 10-4 (c), 1.5 × 10-4 (d), and 2.0 × 10-4 (e). a 151 b 152 c 153 d 154 e 155 Figure 29. Basal melting rate (in mm yr-1) of the DAF ice sheet in simulations with an ice inventory of ~275 m GEL, climate data from a GCM run with a 1000 mb CO2 atmosphere with grey gas κ = 1.0 × 10-4, and geothermal heat flux of (a) 45 mW m-2, (b) 55 mW m-2, and (c) 65 mW m-2. a 156 b 157 c 158 Figure 30. Surface water ice (in kg m-2) after ~400 years of ice evolution in a GCM simulation with 41.8° spin-axis obliquity and 1500 mb CO2 atmosphere (shaded contours), and the areal centroid of the DAF (and putative late Hesperian paleopole location) as calculated by Kite et al. 2006 (black square). 159 Figure 31. Possible mechanisms for generating the meltwater that flowed beneath the DAF ice sheet and allowed eskers to form include (a) basal melting due to elevated geothermal flux, (b) basal melting due to anomalously thick ice, (c) melting due to lava- ice interactions, (d) basal melting due to elevated atmospheric temperatures (“top-down basal melting”), or (e) top-down melting in a substantially warmer climate. a b c d 160 e 161 Figure 32. Possible sites of explosive glaciovolcanism (black arrows) associated with the Dorsa Argentea eskers (white arrows). MOLA topography shading on THEMIS 100 m daytime infrared image mosaic. 162 Chapter 3 Volcanism-Induced, Local Wet-Based Glacial Conditions Recorded in the Late Amazonian Arsia Mons Tropical Mountain Glacier Deposits Kathleen E. Scanlon,1 James W. Head,1 and David R. Marchant2 1 Department of Earth, Environmental and Planetary Sciences, Brown University 324 Brook Street, Box 1846, Providence, RI, 02912, USA 2 Department of Earth and Environment, Boston University 685 Commonwealth Avenue, Boston, MA, 02215, USA Published April 2015 in Icarus, Volume 250, p. 18-31 doi: 10.1016/j.icarus.2014.11.016. 163 Abstract The tropical mountain glacial fan-shaped deposit (FSD) to the northwest of the Arsia Mons volcano on Mars contains numerous glacial and volcanic landforms. While most of the glacial landforms are interpreted to have formed by cold-based glacial processes, several glacial landforms near glaciovolcanic edifices are more consistent with localized wet-based glacial processes. These landforms include ribbed moraines, which suggest local, thermal transitions between wet- and cold-based ice; thrust-block moraines, whose formation is typically assisted by the presence of subglacial water; streamlined knobs that we interpret to have been sculpted by ice sliding along its base; and a braided outflow channel. The presence and association of these features, together with evidence of both subglacial volcanic eruptions and local ice-marginal advances, favor polythermal glaciers with localized wet-based conditions. We propose that lava-to- ice heat transfer during the eruption of the glaciovolcanic edifices caused the Arsia Mons paleoglacier to melt at its base in some areas, resulting in these locally wet-based glacial conditions. A polythermal glacier provides more potential microbial habitats and more connectivity between habitats than does a cold-based glacier, and we review glacial and glaciovolcanic habitats on Earth that may provide insight into the likelihood of potential microbial habitats within the Arsia Mons FSD on Mars. 1. Introduction To the west or northwest of each of the Tharsis Montes lie arcuate deposits, delineated by concentric ridges, known as the lobe- or fan-shaped deposits (e.g. Zimbelman and Edgett, 1992; Scott and Zimbelman, 1995). While the fan-shaped 164 deposits (FSDs) were initially interpreted as having formed by either mass-wasting processes, volcanic processes, or tectonic deformation (see review in Shean et al., 2005), recent work (Head and Marchant, 2003; Shean et al., 2005; Forget et al., 2006; Shean et al., 2007; Kadish et al., 2008; Fastook et al., 2008; Scanlon et al., 2014a) strongly supports the hypothesis (Williams, 1978; Lucchitta, 1981) that the FSDs are glacial in origin. Specifically, the morphology of landforms in the FSDs (Head and Marchant, 2003; Shean et al., 2005; Shean et al., 2007; Kadish et al., 2008; Scanlon et al., 2014a) strongly resemble, and show the same spatial and stratigraphic relationships as, characteristic glacial landforms deposited from cold-based ice on Earth (e.g. Atkins, 2013). The term “cold-based” refers to glaciers whose basal ice temperatures are below the pressure melting point. Results from Mars global climate models show that ice should accumulate on the western flanks of the Tharsis Montes when martian orbital obliquity is high (Forget et al., 2006), and output from coupled ice-sheet climate models closely reproduce the footprints of the deposits (Fastook et al., 2008). Three geomorphologic facies comprise the majority of the deposit at Arsia Mons: a Ridged, Knobby, and Smooth Facies (Zimbelman and Edgett, 1992; Scott and Zimbelman, 1995), interpreted, respectively, as drop moraines, sublimation till, and remnant debris-covered glaciers (Head and Marchant, 2003). Like the other Tharsis Montes FSDs (Shean et al., 2005; Kadish et al., 2008), the Arsia Mons fan-shaped deposit is associated with landforms that have been interpreted as subglacially erupted volcanic edifices (Figure 1). These landforms, outlined in black in Figure 1, include four plateaus interpreted as pillow mounds topped by hyaloclastite mounds; a flat-topped, steep-sided plateau interpreted as a tuya; a steep-sided ridge 165 interpreted as an ice-confined lava flow; a series of digitate cliffs interpreted as ice- contact features; ridges grading into chains of pits, interpreted as eruptions spanning the boundary between permeable and impermeable ice; steep-sided, sunken-centered edifices interpreted as having formed by lava extrusions that chilled at the margins but whose still-molten centers receded into the vent; and numerous low, steep-sided, areally extensive mounds interpreted as pillow mounds (Scanlon et al., 2014a). Amazonian-aged relict glacial deposits, including concentric crater fill, lobate debris aprons and lineated valley fill (e.g. Squyres, 1979; Squyres and Carr, 1986; Mangold, 2003; Milliken et al., 2003; Head et al., 2003; Garvin et al., 2006; Head et al., 2006; Dickson et al. 2008; Head et al., 2010) are found from polar to equatorial regions of Mars. A lack of erosive features (e.g. grooves and molded forms) or evidence for associated fluvial activity, however, suggests that most of these deposits were formed by cold-based glaciers. In the Noachian and Hesperian eras, there is evidence that ice sheets at the south pole (e.g. Head and Pratt, 2001; Fastook et al., 2012), in Valles Marineris (e.g. Gourronc et al., 2014), in Argyre basin (e.g. Banks et al., 2009; Bernhardt et al., 2013), and in outflow channels (e.g. Lucchitta et al., 1981) were wet-based, but as the martian climate became colder, cold-based glaciation became the dominant paradigm. To our knowledge, outside of the Tharsis Montes FSDs, all candidate Amazonian polythermal or wet-based glacial deposits proposed thus far have been restricted to small regions within impact craters (Marchant et al., 2006; Hubbard et al., 2011), where basal melting may have been assisted by remnant thermal gradients after impact (Marchant et al., 2006). 166 On Earth, heat transfer from intrusive or extrusive volcanism underneath an otherwise cold-based ice sheet can be sufficient to allow melting and locally uncouple the glacier from the ground (e.g. Fahnestock et al., 2001; Fox Maule et al., 2005; Head and Wilson, 2007; Hambrey et al., 2008). On Mars, fluvial outflow channels and aligned, streamlined knobs that have been interpreted as drumlins in the Pavonis Mons fan-shaped deposit (Scott et al., 1998; Shean et al., 2007) suggest that volcano-ice interactions may have allowed local basal melting in the Pavonis tropical mountain glacier, though it is unclear whether the stratigraphic position of the streamlined knobs is consistent with the hypothesis that they are drumlins. No evidence for fluvial flow or wet-based glacial conditions has been documented, however, in the Olympus Mons glacial deposits or the Ascraeus Mons FSD (Milkovich et al., 2006; Kadish et al., 2008). In the Arsia Mons FSD, we find several characteristic classes of wet-based glacial landforms. The spatial relationships of these landforms to each other and to the documented glaciovolcanic edifices in the deposit suggest that the Arsia glacier was locally wet-based (i.e., polythermal), with volcanism rather than climate causing the wet-based conditions. 2. Data Using ArcMap 10.0, we constructed a ~5 meters per pixel basemap of all Reconnaissance Orbiter (MRO) Context Camera (CTX) images (Malin et al., 2007) in the Arsia Mons fan-shaped deposit. Where CTX coverage was absent or unclear, we consulted High Resolution Stereo Camera (HRSC) images with 10 – 30 meters per pixel resolution (Neukum and Jaumann, 2004). We obtained topographic information from 128 pixel per degree (~463 meters per pixel) resolution MOLA (Zuber et al., 1992; Smith et 167 al., 1999) data and HRSC-derived Digital Elevation Maps (DEMs) with ~100 meters per pixel resolution (Dumke et al., 2008). These data were processed using the Spatial Analyst toolkit in ArcMap 10.0. 3. Glacial landforms 3.1 Outflow channels, streamlined knobs, and drop moraines A plateau towards the northwest of the Arsia Mons FSD (the “Northwest Plateau”), first described by Wilson and Head (2007a, b) and later by Scanlon et al. (2014), is interpreted to have formed in a subglacial eruption (Figure 2). The broad lower-elevation mass of the plateau is interpreted as an effusive pillow eruption. The steep, central mound superimposed upon it is interpreted in Scanlon et al. (2014) to have formed in a later, explosive phase of the eruption that began after subglacial meltwater drained from the edge of the glacier and the ice above the vent melted and thinned. The following landform associations surrounding the Northwest Plateau are consistent with meltwater production and glacier ice locally decoupling from its base as the volcanic plateau formed (Scanlon et al., 2014a): (1) fluvial channels emerging from underneath glacial debris directly downslope of the plateau (and in the inferred direction of the paleo ice-surface slope), which indicate the production and flow of a substantial volume of water at the base of the glacier; (2) a field of knobs between the plateau and the glacio-fluvial channels that are elongated and streamlined in the downslope and inferred down-ice direction, suggesting local glacial scour; (3) sharp-crested ridges, grading downslope to broad-crested ridges in the inferred ice-flow direction, emerge from the central mound and are interpreted as sinuous accumulations of glacio-fluvial debris 168 (i.e. eskers); and (4) the long axes of the concentric ridges immediately downslope of the plateau are highly arcuate in the inferred down-ice direction, suggesting local increase in longitudinal ice-flow velocity and local ice-margin advance (Figure 4); the non- destructive superposition of these ridges upon other ridges with similar morphology (Figure 3), and the morphological commonalities between these ridges and Ridged Facies members at the same radius from Arsia Mons, suggest that they are drop moraines, and the downslope deflection of their orientation suggests that they were formed at the edges of a lobe of ice elongated in the same direction implied by the other 3 landform classes. 3.2. Ribbed moraines The Lobate Facies (Zimbelman and Edgett, 1992; Figure 1) is a unit in the Arsia Mons FSD characterized by numerous, densely distributed steep-sided, lobate scarps and ridges, some of which grade upslope into chains of pit craters. The steep-sided features are interpreted as subglacial lava flows, and the pit craters as the sites of phreatomagmatic eruptions under thinner ice, suggesting that volcano-ice interactions were common in this region of the FSD (Scanlon et al., 2014a). At the contact between the Lobate and Knobby Facies (Figure 5) lies a ~17 km wide field of sharply defined, commonly asymmetric, regularly spaced, concentric or anastomosing ridges. The ridges are oriented transverse to inferred ice-flow directions, and are up to 30 km long and typically 500 m – 1 km wide (Figure 6). Where HRSC DEM coverage is available, the ridges are typically 10 – 50 m high. On the basis of this morphology, their regular size and spacing, their orientation transverse to inferred paleo ice flow, their alignment with moraines of unusual morphology farther downslope 169 (Section 3.3), and the fact that they do not appear to gently drape underlying topography as do the drop moraines in the Ridged Facies (Shean et al., 2007), we interpret these ridges as ribbed moraines formed in association with glacier ice that locally achieved wet-based conditions (cf. Dunlop and Clark, 2006). Ribbed moraines are unusual and complex glacial landforms. They do not form alongside ice margins, but rather are produced locally at the base of polythermal glaciers (e.g. Dunlop and Clark, 2006 and sources therein; Möller, 2006). The precise mechanism by which they form is incompletely understood, but candidate origins include those in which (1) the moraines are the remnants of a frozen till sheet that extends and fractures when a wet-based, downslope region of a glacier flows more rapidly than a cold-based interior region (Hättestrand and Kleman, 1999); (2) ice-cored moraines left by a cold- based glacier are re-sculpted into ribbed moraines when subsequently overrun by wet- based ice (Möller, 2006; Möller, 2010); (3) the crests of the moraines represent the crests of former waves produced by a fluid dynamical instability in the flow of glacial ice and subglacial till (Dunlop et al., 2008; Fowler, 2009; Chapwanya et al., 2011); or (4) moraines form when seasonal or climatic variations in subglacial meltwater content cause the glacier bed to transition locally from an extensional to a compressional deformation regime (Lindén et al., 2008; Stokes et al., 2008). Whether all subtypes of ribbed moraine form by the same process is debated (e.g. Hättestrand and Kleman, 1999, Finlayson and Bradwell, 2008; Lindén et al., 2008; Chapwanya et al., 2011), but most researchers acknowledge that marked changes in horizontal ice-flow velocity, most easily accomplished beneath polythermal glaciers, are a crucial factor in their formation (Finlayson and Bradwell, 2008). We propose that the heat produced during episodes of 170 extrusive subglacial volcanism in the Lobate Facies led to basal meltwater production, resulting in spatial and temporal gradients in basal sliding and longitudinal ice-flow velocity (Figure 7), consistent with formation mechanisms for ribbed moraines discussed above. Smaller ridges (~200 m wide and not high enough to be resolved by the HRSC DEM), of similar shape and similar spacing relative to their size, occur superimposed upon the Northwest Plateau (Section 3.1; Figure 8). The orientation of these ridges, southwest-to-northeast, is also transverse to the direction of inferred glacial flow based on the orientation of nearby streamlined knobs. Similar to the case described above for the ribbed moraines downslope of the Lobate Facies, we assert that the volcanic construction and subsequent cooling of the Northwest Plateau affected the thermal regime of adjacent glacier ice, producing polythermal conditions that led to the local development of ribbed moraines. The superposition of ridges on top of the plateau suggests an origin via temporal, as well as spatial, transitions between wet-based and cold-based ice conditions. Specifically, the ice directly above the Northwest Plateau would have remained wet-based longer than the ice farther downslope from the volcanic heat source, causing compressive stress and ribbed moraine formation above the plateau. 3.3 Thrust-Block Moraine At the northern edge of the FSD (Figure 1, Figure 9) lies an L-shaped ridge ~50 km long, up to ~450 m high, and up to ~4 km wide. Due to its height, steep sides, and the channeled morphology at its distal end, this ridge is interpreted to have formed as an ice- confined volcanic flow; heat transfer from this volcanic flow to the surrounding ice is 171 conservatively estimated to have generated ~38 km3 of liquid water (Scanlon et al., 2014a). Downslope and to the west of the ridge (Figure 9), the terminal moraine of the FSD differs dramatically in size and appearance from elsewhere in the deposit (Figure 10). The moraine is up to 3 km wide here, whereas it is typically only a few hundred meters wide, and is taller here (up to ~125 m) than elsewhere in the deposit (Figure 11). The moraine has several small parallel ridges at its top and its MOLA profile resolves one broad ridge superimposed on the moraine. Based on its size, surface appearance, and topographic profile, as well as its exclusive occurrence downslope of the region of the FSD containing the largest glaciovolcanic edifice and the highest density of glaciovolcanic landforms, we interpret this feature as a thrust-block moraine (also known as a composite moraine, or sometimes as a push moraine; e.g. Benn and Evans, 2010). The multi-ridged appearance (Figure 10) and profile (Figure 11a, b) of the moraine is consistent with an imbricate stack of thrust slabs. On Earth, such thrust-block moraines form where high longitudinal stresses act on weak subglacial sediments or bedrock, e.g. adjacent to a steeply sloping ice margin or across a thermal boundary between wet-based and cold-based ice conditions. Basal debris, entrained by regelation, may be advected upward along thrust planes in the ice, potentially producing an imbricated moraine ridge. Locally, elevated pore pressures that arise from permafrost-confined aquifers beneath polythermal glaciers may enhance basal deformation and entrainment (e.g. Boulton et al., 1999; Bennett, 2001; Benn and Evans, 2010), leading to the formation of thrust-block moraines. In addition to being influenced 172 by rheological properties of the deforming substrate, the size of thrust-block moraines scale with the magnitude and rate of glacial advance (Bennett, 2001). Within purely cold-based glaciers, thrust-block moraines have only been documented where ice advances into marginal lakes, where wet lacustrine sediment may freeze onto basal ice and move upward along local thrust planes (Fitzsimons, 2003; Hambrey and Fitzsimons, 2010); this process, however, is unlikely to have been the case at Arsia Mons. Instead, we propose that the large moraine ridge in the Arsia Mons FSD reflects polythermal thermal variation near the ice margin. Local, wet-based conditions associated with periods of subglacial volcanism were likely associated with basal entrainment and increased sediment transport downglacier (Figure 12). As the ice flowed from the volcanic heat source it cooled, and saturated basal debris would have frozen onto the cold ice; compressive stresses near the ice margin could then have produced imbricate thrusts, bringing up frozen basal sediment and producing the observed thrust- block moraine. This explanation is bolstered by the fact that the hypothesized thrust- block moraine lies directly downslope of a large field of ribbed moraines (Section 3.3), which form directly over regions postulated to experience transitions from wet-based to cold-based glacier ice. 4. Discussion Four characteristic classes of wet-based or polythermal glacial landforms are observed near glaciovolcanic edifices in the Arsia Mons FSD: (1) outflow channels, which record the release of large volumes of meltwater; (2) streamlined knobs, which record the direction of glacial motion as well as its erosive power; (3) ribbed moraines, 173 which are thought to indicate transitions between wet-based and cold-based glacier ice; and (4) a large thrust-block moraine, which records the local deformation of basal sediment by large glaciotectonic stresses typically associated with either fast-flowing glacier ice (surging) or transitions from wet-based to cold-based glacier ice. Together, the landforms provide a coherent history at each glaciovolcanic site (Figure 13): uniform ice- flow directions are recorded both by the streamlined knobs and the ribbed moraines at the Northwest Plateau. Furthermore, the outflow channels emerge directly beyond these landforms, as would be expected if the enhanced flow was stimulated by wet-based conditions. Similarly, the ribbed moraines downslope of the numerous glaciovolcanic edifices in the Lobate Facies and the thrust-block moraine farther downslope at the terminus of the deposit both point to a transition from wet-based to cold-based ice, which is consistent with the placement of glaciovolcanic landforms in the FSD (Figure 13). These consistent landform associations imply that glaciovolcanic activity was sufficient to alter the local thermal regime of the Arsia Mons tropical mountain glacier, and their exclusive association with glaciovolcanic edifices implies that climate and pressure effects were never by themselves sufficient to cause basal melting in the glacier. 5. Potential microbial habitats in the Arsia Mons FSD On Earth, glaciovolcanic environments provide a range of microbial habitats (e.g. Cousins and Crawford, 2011). While the chemical conditions would have differed from those in terrestrial glaciers due to the lack of an oxygen atmosphere and a vigorous biosphere, plausible redox couples and nutrient sources could have made wet glacial environments on Mars hospitable to life (Cockell et al., 2011). Since such environments 174 would have originated from volcano-ice interaction in the case of this deposit, volcanic gases in some of the habitats would have further strengthened redox gradients and provided an additional carbon source in the form of volcanic CO2 (Skidmore, 2011). The following environments, which geomorphological evidence indicates were present in the Arsia Mons FSD paleoglacier, are inhabited on Earth: 5.1 Englacial and basal lakes. The glaciovolcanic landforms in the Arsia Mons FSD are likely to have generated several englacial lakes with volumes on the order of tens of cubic kilometers and longevity of hundreds to thousands of years (Scanlon et al., 2014a). The few terrestrial volcanically generated englacial lakes whose ecosystems have been surveyed (Gaidos et al., 2004; Gaidos et al., 2008; Marteinsson et al., 2013) contained microbial communities capable of fixing carbon and nitrogen, and including organisms related to known acetogens, sulfate reducers and iron reducers. These communities are supported by sediments and bottom waters rich in volcanic CO2, H2, ferric iron, and sulfur compounds, as well as oxygen from the melted glacial ice (Jóhannesson et al., 2007; Marteinsson et al., 2013). At least one basal lake beneath 800 m thick ice in the Antarctic Ice Sheet has been shown to host a microbial community supported by chemoautotrophic production (Christner et al., 2014). Since the Arsia Mons paleoglacier experienced transient wet- based conditions, analogous basal lakes within the FSD may have provided additional habitats for microbial life. Other ice-covered lakes, such as those in the Antarctic Dry Valleys (e.g. Roberts et al., 2000; Murray et al., 2012), are poorer analogs to the Arsia Mons lakes because they originated as subaerial lakes. Subaerial lakes can be colonized more easily than englacial lakes due to their exposure to surface runoff and dormant, 175 windborne microorganisms, and after they develop an ice cover their productivity can still be boosted by legacy carbon from the earlier lake ecosystem and other nearby environments (Priscu et al., 1999). Phylogenetic evidence suggests that the microbes that colonize terrestrial subglacial volcanic lakes originate in a variety of other environments, including glacial ice and snow, dust and marine aerosols that work their way into the lakes long after being deposited on the glacier surface, drainage of surface meltwater into the lake (Gaidos et al., 2004), or subsurface aquifers in contact with the subglacial melt (Marteinsson et al., 2013). Neither surface runoff nor iced-over subaerial lakes are likely to have been present near the subglacial lakes in the Arsia Mons FSD. The dike intrusions in the Arsia Mons FSD could potentially have warmed the subsurface enough and melted enough ground ice for a liquid aquifer to persist (e.g. Head and Wilson, 2002). Such an aquifer could have hydrologically connected some habitats within the deposit, enabling microorganisms to move between them. However, due to the regional dust cover mantling the deposits, no spectral evidence for hydrothermal alteration minerals has yet been found in the deposit, and any fine-scale morphologic features of hydrothermal systems would have been obscured by glacial debris in the Ridged, Knobby and Smooth Facies. If circulating hydrothermal fluid was not present, putative martian life could still have colonized the lakes if airborne propagules (perhaps the long-dormant inhabitants of another transient surface habitat) worked their way into the glacial ice, or if the ice itself was inhabited or contained dormant life forms. 5.2 Wet-based glacier ice. While the Arsia Mons glacier was cold-based through much of its spatial and temporal extent (Head and Marchant, 2003), we have shown that 176 subglacial eruptions induced local wet-based conditions in several regions of the Arsia Mons paleoglacier. Microbial communities have been observed directly (e.g. by microscopy or metagenomic analyses) or indirectly (e.g. by isotopic measurements) in the basal ice, water, brines, and sediments of wet-based and polythermal glaciers in Alaska, Antarctica, Canada, Greenland, New Zealand, and Norway (see review in Hodson et al., 2008). In some cases these communities are more diverse than the surface glacier communities (Hamilton et al., 2013). The environment appears to support organisms that participate in iron and sulfur cycling (Mikucki et al., 2004; Mikucki et al., 2009), methanogenesis (Stibal et al., 2012a), and other processes that could have been viable on Mars. Wet subglacial sediments in the Arsia Mons glacier could have been inoculated by the same sources as subglacial lakes: groundwater connections, organisms in the ice, or organisms in surficial or englacial debris. 5.3 Glacial ice and debris. The ubiquity of the Knobby Facies, which forms from the sublimation of debris-rich ice (Head and Marchant, 2003), indicates that much of the Arsia Mons glacier was rich in debris. In terrestrial glaciers, debris-rich ice contains more microorganisms than clean ice because the debris is a source of nutrients and electron donors or acceptors. Furthermore, thin films of liquid water are present at mineral-ice interfaces at temperatures well below freezing (Wettlaufer, 1999; Price, 2007). Even when these films are too small for microbes to move freely, dissolved ions and organic compounds are still mobile and allow microbes to carry out metabolic processes including methanogenesis, denitrification, sulfate reduction, and iron reduction (see review in Price, 2007). 177 Clean ice can also contain habitable microenvironments at subfreezing temperatures. Small amounts of impurities (including nutrients and organic compounds) have been shown to segregate into veins between grains of ice due to their low solubility in solid ice. These veins contain concentrated nutrients and are large enough to allow microbes to move within them, and are thought (Price, 2000) to be the reason viable microbes have often been isolated from ice cores (e.g. Christner et al., 2000). Organisms adapted to subfreezing temperatures can survive, perhaps indefinitely (Price, 2000), and even grow in these veins (Mader et al., 2006; Dani et al., 2012). While such environments are unlikely to be as productive as macroscopic englacial water bodies, they may have represented refugia from which these larger bodies could be colonized. Remnant ice in the deposit (Head and Marchant, 2003; Shean et al., 2007; Scanlon et al., 2014b), could provide these habitats even today. Hollowed ridges at the western edge of the deposit (Scanlon et al., 2014b), which represent former snow dunes, indicate that snow or ice was transported and reworked by wind in the region surrounding the deposit. This would have helped disperse any ice-dwelling organisms, increasing the potential for colonization of any glacial or glaciovolcanic habitats in the region. 5.4 Hyaloclastite and palagonite. While the subglacially erupted edifices in the deposit have not yet been imaged with resolution high enough to identify surface textures, and the material mantling the deposit may prevent confident identification even then, these structures should all contain abundant basaltic glass due to the rapid chilling of basaltic lava against ice as it erupted. We have argued (Scanlon et al., 2014a) that many of the landforms within the FSD should have glassy surfaces, and that several 178 should consist largely of rapidly quenched basaltic tephra. Compared to crystalline basalt, hyaloclastite is a favorable habitat for microbial life (as is its alteration product, palagonite) because the lack of crystal structure makes nutrients in the rock more easily accessible (Cockell et al., 2009). These basaltic glasses are inhabited in exposed flows that were erupted subglacially, in submarine lavas, and in subglacial lakes (e.g. Cockell et al., 2009; Gaidos et al., 2004; and Thorseth et al., 2001, respectively), and are among the environments in which putative evidence of early terrestrial life has been found (Banerjee et al., 2006). The range of autotrophic metabolisms supported by the volcanic glass environment on Earth remains incompletely characterized, however (Cockell, 2011). 5.5 Biosignature preservation potential. In environments well-suited to the preservation of molecular and morphological biosignatures, biogenic material is concentrated by physical processes, protected by rapid burial from oxidation and ultraviolet radiation at the surface, and isolated from aqueous or thermal alteration after burial; ideally, phyllosilicates are deposited simultaneously, aiding in preservation (Summons et al., 2011). While spectrometers have thus far been unable to conclusively determine the mineralogy of the region, several landforms in the deposit meet the other criteria. The Northwest Plateau is one example of a location within the FSD where biosignatures may have been generated and preserved. In terrestrial subglacial volcanic lakes, microbes are concentrated in the hyaloclastite bottom sediments rather than in the water column (Gaidos et al., 2004). Any microorganisms inhabiting the englacial lake above the Northwest Plateau would almost certainly have been concentrated in the hyaloclastite for the same reasons. Some of these sediments currently comprise the eskers 179 extending from the tephra mound; the top few meters of sediment in the larger esker would have easily been able to protect biogenic material deeper in the esker against ionizing radiation (Kminek and Bada, 2006). The rest of the sediments are similarly protected by meters of Knobby Facies material. Furthermore, the knobby material itself is former englacial debris; as the Arsia Mons glacier sublimed, any organic material that froze out of the water column in the lake would ultimately be concentrated in the Knobby Facies, along with material from habitats that may have existed in the Arsia Mons glacier but are too small to detect by remote sensing, such as the sub-meter-scale “cryoconite holes” melted into ice where low-albedo supraglacial debris aggregates (e.g. Stibal et al., 2012b). There is no evidence to suggest that the eskers, the central mound of the plateau, or the Knobby Facies have been fluvially or thermally altered since their formation, except possibly by the deposition of volcanic ash. 6. Conclusions Glaciovolcanic landforms are abundant in the Arsia Mons fan-shaped deposit. Previous work has made a strong case for cold-based glaciation at the Tharsis Montes, but new, higher-resolution data show that landforms indicative of local wet-based conditions due to interaction between lava and glacial ice are also distributed throughout the deposit. These landforms, which record local increases in ice-flow velocity and polythermal conditions, include thrust-block moraines, ribbed moraines, highly arcuate drop moraines, streamlined knobs, and fluvial outflow channels. Since these landforms all occur exclusively in association with glaciovolcanic features, we conclude that the 180 Arsia Mons FSD was left by a glacier whose local wet-based conditions were induced by volcanic warming rather than by climate. The aqueous environments evidenced by the Arsia Mons FSD are exceptional in that they are comparable in size to similar Noachian-Hesperian environments, and potentially much longer-lived than other Amazonian environments. Furthermore, the Arsia Mons glaciovolcanic environments occurred recently enough (Kadish et al., 2014) that unambiguous molecular biomarkers have been retrieved from much older terrestrial environments (e.g. Brocks et al., 1999). If Mars Science Laboratory (Grotzinger et al., 2012) or other upcoming missions discover evidence for fossil or extant surface life on Mars, the Arsia Mons Fan-Shaped Deposit would be a well-suited target for future human or robotic exploration to determine whether it persisted into the modern era. Acknowledgements We gratefully acknowledge support from the NASA Graduate Student Researchers Program (Grant NNX12AI39H) to KES, from the Mars Data Analysis Program (Grant NNX11AI81G) and the Mars Express High-Resolution Stereo Camera (HRSC) Investigation Team (JPL 1488322) to JWH, and from NSF Polar Programs (Grant ANT- 0944702) to DRM. References Atkins, C.B. 2013. Geomorphological evidence of cold-based glacier activity in South Victoria Land, Antarctica. In Hambrey, M. J., Barker, P. F., Barrett, P. J., 181 Bowman, V., Davies, B., Smellie, J. L. & Tranter, M. (eds) 2013. Antarctic Palaeoenvironments and Earth-Surface Processes. Geological Society, London, Special Publications, 381, 299–318. First published online July 24, 2013, http://dx.doi.org/10.1144/SP381.18 Banerjee, N. R., H. Furnes, K. Muehlenbachs, H. Staudigel, and M. de Wit (2006), Preservation of 3.4–3.5 Ga microbial biomarkers in pillow lavas and hyaloclastites from the Barberton Greenstone Belt, South Africa, Earth and Planetary Science Letters, 241(3-4), 707–722. Banks, M. E., N. P. Lang, J. S. Kargel, A. S. McEwen, V. R. Baker, J. A. Grant, J. D. Pelletier and R. G. Strom (2009), An analysis of sinuous ridges in the southern Argyre Planitia, Mars using HiRISE and CTX images and MOLA data, Journal of Geophysical Research: Planets, 114 (E9). Benn, D. I., and D. J. Evans (2010), Glaciers and Glaciation, 2nd edition. 816 p. Hodder Education. Bennett, M. R. (2001), The morphology, structural evolution and significance of push moraines, Earth-Science Reviews, 53(3-4), 197–236. Bernhardt, H., H. Hiesinger, D. Reiss, M. Ivanov, G. Erkeling (2013), Putative eskers and new insights into glacio-fluvial depositional settings in southern Argyre Planitia, Mars, Planetary and Space Science, 85, 261–278. Boulton, G. S., J. J. M. van der Meer, D. J. Beets, J. K. Hart, and G. H. J. Ruegg (1999), The sedimentary and structural evolution of a recent push moraine complex: Holmstrømbreen, Spitsbergen, Quaternary Science Reviews, 18(3), 339—371. 182 Brocks, J. J., G. A. Logan, R. Buick, and R. E. Summons (1999), Archean molecular fossils and the early rise of eukaryotes, Science, 285(5430), 1033-1036. Chapwanya, M., C. D. Clark, and A. C. Fowler (2011), Numerical computations of a theoretical model of ribbed moraine formation, Earth Surface Processes and Landforms, 36(8), 1105–1112. Christner, B. C., E. Mosley-Thompson, L. G. Thompson, V. Zagorodnov, K. Sandman and J. N. Reeve (2000), Recovery and Identification of Viable Bacteria Immured in Glacial Ice, Icarus, 144, 479–485. Christner, B.C., et al. (2014). A microbial ecosystem beneath the West Antarctic ice sheet, Nature, 512, 310–313. Cockell, C. S. (2011), Life in the lithosphere, kinetics and the prospects for life elsewhere, Philosophical Transactions of The Royal Society A, 369(1936), 516- 537. Cockell, C. S., K. Olsson, F. Knowles, L. Kelly, A. Herrera, T. Thorsteinsson, and V. Marteinsson (2009), Bacteria in Weathered Basaltic Glass, Iceland, Geomicrobiology Journal., 26(7), 491-507. Cockell, C. S., E. Bagshaw, M. Balme, P. Doran, C. P. McKay, K. Miljkovic, D. Pearce, M. J. Siegert, M. Tranter, and M. Voytek (2011), Subglacial environments and the search for life beyond the Earth, in Antarctic Subglacial Aquatic Environments, Geophysical Monograph Series, 192, eds. M. J. Siegert, M. C. Kennicutt II, and R. A. Bindschadler, 129-148. American Geophysical Union. Cousins, C. R., and I. A. Crawford (2011), Volcano-ice interaction as a microbial habitat on Earth and Mars, Astrobiology, 11(7), 695-710. 183 Dani, K. G. S., H. M. Mader, E. W. Wolff, and J. L. Wadham (2012), Modelling the liquid-water vein system within polar ice sheets as a potential microbial habitat, Earth and Planetary Science Letters, 333, 238-249. Dickson, J. L., J. W. Head, and D. R. Marchant (2008), Late Amazonian glaciation at the dichotomy boundary on Mars: Evidence for glacial thickness maxima and multiple glacial phases, Geology, 36, 411-414. Dumke, A., M. Spiegel, R. Schmidt, and G. Neukum (2008), High-resolution digital terrain models and ortho-image mosaics of Mars: Generation on the basis of Mars-Express HRSC data, presented at Lunar and Planetary Science Conference, XXXIX, abstract #1910. Dunlop, P., and C. D. Clark (2006), The morphological characteristics of ribbed moraine, Quaternary Science Reviews, 25(13-14), 1668–1691. Dunlop, P., C. D. Clark, and R. C. A. Hindmarsh (2008), Bed Ribbing Instability Explanation: Testing a numerical model of ribbed moraine formation arising from coupled flow of ice and subglacial sediment, Journal of Geophysical Research: Earth Surface, 113(F03). Fahnestock, M., W. Abdalati, I. Joughin, J. Brozena, and P. Gogineni (2001), High Geothermal Heat Flow, Basal Melt, and the Origin of Rapid Ice Flow in Central Greenland, Science, 294(5550), 2338-2342. Fastook, J. L., J. W. Head, D. R. Marchant, and F. Forget (2008), Tropical mountain glaciers on Mars: Altitude-dependence of ice accumulation, accumulation conditions, formation times, glacier dynamics, and implications for planetary spin-axis/orbital history, Icarus, 198(2), 305-317. 184 Fastook, J. L., J. W. Head III, D. R. Marchant, F. Forget, and J.-B. Madeleine (2012), Early Mars climate near the Noachian-Hesperian boundary: Independent evidence for cold conditions from basal melting of the south polar ice sheet (Dorsa Argentea Formation) and implications for valley network formation, Icarus, 219, 25-40. Finlayson, A. G., and T. Bradwell (2008), Morphological characteristics, formation and glaciological significance of Rogen moraine in northern Scotland, Geomorphology, 101(4), 607–617. Fitzsimons, S. J. (2003), Ice-Marginal Terrestrial Landsystems: Polar Continental Glacier Margins, in Glacial Landsystems, ed. D. Evans. Routledge, London. Forget, F., R. Haberle, F. Montmessin, B. Levrard, and J. Head (2006), Formation of glaciers on Mars by atmospheric precipitation at high obliquity, Science, 311(5759), 368-371. Fowler, A. C. (2009), Instability modelling of drumlin formation incorporating lee-side cavity growth, Proceedings of the Royal Society A, 465(2109), 2681-2702. Fox Maule, C., M. E. Purucker, N. Olsen, and K. Mosegaard (2005), Heat flux anomalies in Antarctica revealed by satellite magnetic data, Science, 309(5733), 464–467. Gaidos, E., B. Lanoil, T. Thorsteinsson, A. Graham, M. Skidmore, S.-K. Han, T. Rust, and B. Popp (2004), A viable microbial community in a subglacial volcanic crater lake, Iceland, Astrobiology, 4(3), 327-344. Gaidos, E., V. Marteinsson, T. Thorsteinsson, T. Jóhannesson, Á. R. Rúnarsson, A. Stefansson, B. Glazer, B. Lanoil, M. Skidmore, and S. Han (2008), An oligarchic 185 microbial assemblage in the anoxic bottom waters of a volcanic subglacial lake, The ISME journal, 3(4), 486-497. Garvin, J. B., J. W. Head, D. R. Marchant, and M. A. Kreslavsky (2006), High-latitude cold-based glacial deposits on Mars: Multiple superposed drop moraines in a crater interior at 70°N latitude, Meteoritics & Planetary Science, 41(10), 1659– 1674. Gourronc, M, O. Bourgeois, D. Mège, S. Pochat, B. Bultel, M. Massé, L. Le Deit, S. Le Mouélic, and D. Mercier (2014), One million cubic kilometers of fossil ice in Valles Marineris: Relicts of a 3.5 Gy old glacial landsystem along the Martian equator, Geomorphology, 204, 235–255. Grotzinger, J. P., et al. (2012), Mars Science Laboratory Mission and Science Investigation, Space Science Reviews, 170, 5–56. Hambrey, M. J., J. L. Smellie, A. E. Nelson, and J. S. Johnson (2008), Late Cenozoic glacier-volcano interaction on James Ross Island and adjacent areas, Antarctic Peninsula region, Geological Society of America Bulletin, 120(5-6), 709-731. Hambrey, M. J., and S. J. Fitzsimons (2010), Development of sediment–landform associations at cold glacier margins, Dry Valleys, Antarctica, Sedimentology, 57(3), 857–882. Hamilton, T. L., J. W. Peters, M. L. Skidmore, and E. S. Boyd (2013), Molecular evidence for an active endogenous microbiome beneath glacial ice, The ISME Journal, 7, 1402–1412 Hättestrand, C., and J. Kleman (1999), Ribbed moraine formation, Quaternary Science Reviews, 18(1), 43–61. 186 Head, J. W., and S. F. Pratt (2001), Extensive Hesperian-aged south polar ice sheet on Mars: Evidence for massive melting and retreat, and lateral flow and ponding of meltwater, Journal of Geophysical Research: Planets, 106(E6), 12,275-12,299. Head, J. W., and L. Wilson (2002), Mars: A review and synthesis of general environments and geological settings of magma—H2O interactions, in Volcano– Ice Interaction on Earth and Mars, Geological Society, London, Special Publication, eds. J. L. Smellie and M. G. Chapman, The Geological Society of London, London. Head, J. W., and D. R. Marchant (2003), Cold-based mountain glaciers on Mars: Western Arsia Mons, Geology, 31(7), 641-644. Head, J. W., D. R. Marchant, M. C. Agnew, C. I. Fassett, and M. A. Kreslavsky (2006), Extensive valley glacier deposits in the northern mid-latitudes of Mars: Evidence for Late Amazonian obliquity-driven climate change, Earth and Planetary Science Letters, 241, 663-671. Head, J. W., and L. Wilson (2007), Heat transfer in volcanoice interactions on Mars: synthesis of environments and implications for processes and landforms, Annals of Glaciology, 45(1), 1-13. Head, J. W., D. R. Marchant, J. L. Dickson, A. M. Kress, and D. M. H. Baker (2010), Northern mid-latitude glaciation in the Late Amazonian period of Mars: Criteria for the recognition of debris-covered glacier and valley glacier landsystem deposits, Earth and Planetary Science Letters, 294, 306-320. 187 Hodson, A., A. M. Anesio, M. Tranter, A. Fountain, M. Osborn, J. Priscu, J. Laybourn- Parry, and B. Sattler (2008), Glacial ecosystems, Ecological Monographs, 78(1), 41-67. Hubbard, B., R. E. Milliken, J. S. Kargel, A. Limaye, C. Souness (2011), Geomorphological characterisation and interpretation of a mid-latitude glacier- like form: Hellas Planitia, Mars, Icarus, 211(1), 330–346. Jóhannesson, T., T. Thorsteinsson, A. Stefánsson, E. J. Gaidos, and B. Einarsson (2007), Circulation and thermodynamics in a subglacial geothermal lake under the Western Skaftá cauldron of the Vatnajökull ice cap, Iceland, Geophysical Research Letters, 34(19), L19502. Kadish, S. J., J. W. Head, R. L. Parsons, and D. R. Marchant (2008), The Ascraeus Mons fan-shaped deposit: Volcano-ice interactions and the climatic implications of cold-based tropical mountain glaciation, Icarus, 197(1), 84-109. Kadish, S. J., J. W. Head III, J. L. Fastook, and D. R. Marchant (2014), Middle to Late Amazonian tropical mountain glaciers on Mars: The ages of the Tharsis Montes fan-shaped deposits, Planetary and Space Science, 91, 52-59. Kminek, G., and J. L. Bada (2006), The effect of ionizing radiation on the preservation of amino acids on Mars, Earth and Planetary Science Letters, 245(1-2), 1-5. Lindén, M., P. Möller, and L. Adrielsson (2008), Ribbed moraine formed by subglacial folding, thrust stacking and lee-side cavity infill, Boreas, 37(1), 102–131. Lucchitta, B. K. (1981), Mars and Earth - Comparison of Cold-Climate Features, Icarus, 45(2), 264-303. 188 Lucchitta, B. K., D. M. Anderson, and H. Shoji (1981), Did ice streams carve martian outflow channels?, Nature, 290, 759 - 763. Mader, H. M., M. E. Pettitt, J. L. Wadham, E. W. Wolff, and R. J. Parkes (2006), Subsurface ice as a microbial habitat, Geology, 34(3), 169-172. Malin, M. C., et al. (2007), Context Camera Investigation on board the Mars Reconnaissance Orbiter, Journal of Geophysical Research: Planets, 112(E5). Mangold, N. (2003), Geomorphic analysis of lobate debris aprons on Mars at Mars Orbiter Camera scale: Evidence for ice sublimation initiated by fractures, Journal of Geophysical Research: Planets, 108(E4). Marchant, D. R., J. W. Head, and M. A. Kreslavsky (2006), Mid-Latitude Glacial Modification of Moreux Crater (44°E, 42°N; 135 km): Evidence for Polythermal Glaciation Related to Impact-induced Enhanced Thermal Gradients, presented at Lunar and Planetary Science Conference XXXVII, abstract 1425. Marteinsson, V. T., Á. Rúnarsson, A. Stefánsson, T. Thorsteinsson, T. Jóhannesson, S. H. Magnússon, E. Reynisson, B. Einarsson, N. Wade, H. G. Morrison, and E. Gaidos (2013), Microbial communities in the subglacial waters of the Vatnajokull ice cap, Iceland, The ISME Journal, 7(2), 427-437. Mikucki, J. A., C. M. Foreman, B. Sattler, W. B. Lyons, and J. C. Priscu (2004), Geomicrobiology of Blood Falls: an iron-rich saline discharge at the terminus of the Taylor Glacier, Antarctica, Aquatic Geochemistry, 10(3-4), 199-220. Mikucki, J. A., A. Pearson, D. T. Johnston, A. V. Turchyn, J. Farquhar, D. P. Schrag, A. D. Anbar, J. C. Priscu, and P. A. Lee (2009), A contemporary microbially maintained subglacial ferrous" ocean", Science, 324(5925), 397-400. 189 Milkovich, S. M., J. W. Head, and D. R. Marchant (2006), Debris-covered piedmont glacier deposits along the northwest flank of the Olympus Mons scarp: Evidence for low-latitude ice accumulation during the Late Amazonian of Mars, Icarus, 181, 388-407. Milliken, R. E., J. F. Mustard and D. L. Goldsby (2003), Viscous flow features on the surface of Mars: Observations from high-resolution Mars Orbiter Camera (MOC) images, Journal of Geophysical Research: Planets, 108(E6). Möller, P. (2006), Rogen moraine: an example of glacial reshaping of pre-existing landforms, Quaternary Science Reviews, 25(3-4), 362–389. Möller, P. (2010), Melt-out till and ribbed moraine formation, a case study from south Sweden, Sedimentary Geology, 232(3-4), 161–180. Murray, A. E., F. Kenig, C. H. Fritsen, C. P. McKay, K. M. Cawley, R. Edwards, E. Kuhn, D. M. McKnight, N. E. Ostrom, and V. Peng (2012), Microbial life at −13° C in the brine of an ice-sealed Antarctic lake, Proceedings of the National Academy of Sciences, 109(50), 20626-20631. Neukum, G., and R. Jaumann (2004), HRSC: The High Resolution Stereo Camera of Mars Express, in Mars Express: The Scientific Payload, ed. A. Wilson, 17-35. European Space Agency. Price, P. B. (2000), A habitat for psychrophiles in deep Antarctic ice, Proceedings of the National Academy of Sciences, 97(3), 1247-1251. Price, P. B. (2007), Microbial life in glacial ice and implications for a cold origin of life, FEMS Microbiology Ecology, 59(2), 217-231. 190 Priscu, J. C., C. F. Wolf, C. D. Takacs, C. H. Fritsen, J. Laybourn-Parry, E. C. Roberts, B. Sattler, and W. B. Lyons (1999), Carbon transformations in a perennially ice- covered Antarctic lake, Bioscience, 49(12), 997-1008. Roberts, E. C., J. Laybourn‐Parry, D. M. McKnight, and G. Novarino (2000), Stratification and dynamics of microbial loop communities in Lake Fryxell, Antarctica, Freshwater Biology, 44(4), 649-661. Scanlon, K. E., J. W. Head, L. Wilson, and D. R. Marchant (2014a), Volcano-Ice Interactions in the Arsia Mons Tropical Mountain Glacier Deposits, Icarus, 237, 315-339. Scanlon, K. E., J. W. Head, L. Wilson, and D. R. Marchant (2014b), Remnant Buried Ice in the Equatorial Regions of Mars: Morphological Indicators Associated with the Arsia Mons Tropical Mountain Glacier Deposits, in preparation. Scott, D., and J. Zimbelman (1995), Geologic map of Arsia Mons Volcano, Mars, U. S. Geological Survey Miscellaneous Investigations Series, Map I-2480. Scott, D. H., J. M. Dohm, and J. R. Zimbelman (1998), Geologic map of Pavonis Mons Volcano, Mars, U.S. Geological Survey Miscellaneous Investigations Series, Map I-2561. Shean, D. E., J. W. Head, and D. R. Marchant (2005), Origin and evolution of a cold- based tropical mountain glacier on Mars: The Pavonis Mons fan-shaped deposit, Journal of Geophysical Research: Planets, 110(E5). Shean, D. E., J. W. Head, J. L. Fastook, and D. R. Marchant (2007), Recent glaciation at high elevations on Arsia Mons, Mars: Implications for the formation and 191 evolution of large tropical mountain glaciers, Journal of Geophysical Research: Planets, 112(E3). Skidmore, M. (2011), Microbial communities in Antarctic subglacial aquatic environments, Geophysical Monograph Series, 192, 61-81. Smith, D. E., M. T. Zuber, S. C. Solomon, R. J. Phillips, J. W. Head, J. B. Garvin, W. B. Banerdt, D. O. Muhleman, G. H. Pettengill, G. A. Neumann, F. G. Lemoine, J. B. Abshire, O. Aharonson, C. D. Brown, S. A. Hauck, A. B. Ivanov, P. J. McGovern, H. J. Zwally, and T. C. Duxbury (1999), The Global Topography of Mars and Implications for Surface Evolution, Science, 284(5419), 1495-1503. Squyres, S. W. (1979), The distribution of lobate debris aprons and similar flows on Mars, Journal of Geophysical Research: Solid Earth, 84(B14), 8087–8096. Squyres, S. W., and M. H. Carr (1986), Geomorphic Evidence for the Distribution of Ground Ice on Mars, Science, 231(4735), 249-252. Stibal, M., J. L. Wadham, G. P. Lis, J. Telling, R. D. Pancost, A. Dubnick, M. J. Sharp, E. C. Lawson, C. E. Butler, and F. Hasan (2012a), Methanogenic potential of Arctic and Antarctic subglacial environments with contrasting organic carbon sources, Global Change Biology, 18(11), 3332-3345. Stibal, M., M. Šabacká, and J. Žárský (2012b), Biological processes on glacier and ice sheet surfaces, Nature Geoscience, 5(11), 771-774. Stokes, C.R., O. B. Lian, S. Tulaczyk, and C. D. Clark (2008), Superimposition of ribbed moraines on a palaeo-ice-stream bed: implications for ice stream dynamics and shutdown, Earth Surface Processes and Landforms, 33(4), 593–609. 192 Summons, R. E., J. P. Amend, D. Bish, R. Buick, G. D. Cody, D. J. Des Marais, G. Dromart, J. L. Eigenbrode, A. H. Knoll, and D. Y. Sumner (2011), Preservation of Martian Organic and Environmental Records: Final Report of the Mars Biosignature Working Group, Astrobiology, 11(2), 157-181. Thorseth, I., T. Torsvik, V. Torsvik, and F. Daae (2001), Diversity of life in ocean floor basalt, Earth and Planetary Science Letters, 194(1), 31-37. Wettlaufer, J. S. (1999), Impurity Effects in the Premelting of Ice, Physical Review Letters, 82(12), 2516-2519. Williams, R. (1978), Geomorphic processes in Iceland and on Mars: A comparative appraisal from orbital images, Geological Society of America Abstracts with Programs, 10, 517. Wilson, L., and J. W. Head (2007a), Arsia Mons Cold-Based Tropical Mountain Glacier: Subglacial Eruptions, Polythermal Glaciation, and Distal Drainage of Meltwater, presented at Volcano-Ice Interactions 2. Wilson, L., and J. W. Head (2007b), Tropical Mountain Glaciers on Mars: Volcano-Ice Interactions and Influences on Glacial Evolution, presented at 7th International Conference on Mars. Zimbelman, J. R., and K. S. Edgett (1992), The Tharsis-Montes, Mars - Comparison of Volcanic and Modified Landforms, presented at Lunar and Planetary Science Conference, XXII, p. 31-44. Zuber, M. T., D. E. Smith, S. C. Solomon, D. O. Muhleman, J. W. Head, J. B. Garvin, J. B. Abshire, and J. L. Bufton (1992), The Mars-Observer Laser Altimeter Investigation, Journal of Geophysical Research: Planets, 97(E5). 193 Figure 1. Geomorphological unit map of the Arsia Mons fan-shaped deposit (FSD), after Zimbelman and Edgett (1992) and Scott and Zimbelman (1995). Legend for color swaths is provided in figure. Elsewhere, thin red lines denote large volcanic grabens, white lines denote contacts between units (dashed where inferred), and black lines denote the outlines of glaciovolcanic landforms and modified moraine ridges. Blue and white dots in the smooth facies denote pits and knobs, respectively. THEMIS 100 m / pixel daytime image mosaic; modified from Scanlon et al. (2014). 194 Figure 2. The Northwest Plateau: (a) CTX image mosaic. (b) Sketch map. The Ridged Facies is shaded in red and the Knobby Facies is shaded in blue. Fluvial channels are denoted by light blue lines and streamlined knobs are denoted by green lines. Boxed insets show the locations of Figures 3 and 8. Modified from Scanlon et al. (2014). a 195 b 196 Figure 3. Close view of drop moraines downslope of the Northwest Plateau. Drop moraines reflecting locally enhanced glacial flow (in the direction indicated by cyan arrow) are non-erosively superimposed (red arrows) upon earlier drop moraines concentric with the boundaries of the Ridged Facies. CTX image P19_008605_1772_XI_02S129W. 197 Figure 4. Map view of volcanic eruption (bright orange/gray) under ice sheet (white) on flank of volcano (dark brown), and resulting moraines (light brown lines): (a and b) Local eruption generates heat and initiates ice acceleration and ice-margin advance. (c) As volcanism wanes, basal ice cools below the pressure melting point and subglacial materials may be entrained; the entrained debris, as well as any previously deposited supraglacial debris, eventually falls passively at stationary ice margins, producing drop moraines. (d and e) As the glacier recedes, drop moraines left behind will continue to delineate local still stands in overall ice recession. Based on variable ice margin geometry, these superposed moraines may cross-cut, at oblique angles, drop moraines deposited during earlier times (Fig. 3). a b c d e 198 Figure 5. Regional context map showing contacts between the Lobate, Ridged, and Knobby facies. The area shown in Figure 6 is highlighted by a white box. 199 Figure 6. Concentric, sharp, evenly spaced ridges, interpreted as ribbed moraines, at the contact between the Lobate Facies and the Knobby Facies (location indicated by white box in Figure 5). CTX image mosaic. 200 Figure 7. Ribbed moraines may form where a glacier transitions from wet-based to cold- based conditions, either over time or along a spatial gradient. The field of large ribbed moraines in the Arsia Mons fan-shaped deposit may have formed where subglacial eruptions in the Lobate Facies caused a transition from wet-based conditions upglacier to cold-based conditions downglacier. 201 Figure 8. Ridges interpreted as ribbed moraines (highlighted by white arrows) on the Northwest Plateau (location indicated in Figure 2b). CTX image mosaic. 202 Figure 9. The L-shaped ridge and surrounding landforms. (a) THEMIS daytime IR image mosaic. (b) Sketch map of the region. a b 203 Figure 10. Terminal moraine near the L-shaped ridge, showing ribbed surface appearance (red arrows). CTX image mosaic. 204 Figure 11. (a) Topographic profile across the line A – A’ in Figure 9b, showing the increased height and ridged surface associated with inferred glaciotectonism. (b) Topographic profile across the line B – B’ in Figure 9b, showing the increased height and ridged surface associated with inferred glaciotectonism. (c) Topographic profile across the line C – C’ in Figure 9b, show a typical profile across the terminal moraine outside the region of wet-based glacial thrusting. Note the difference in y-axis scale between the three plots. Topographic data from HRSC-derived DEM. a 205 b c 206 Figure 12. We propose that the enlarged, ridged moraine at the northern edge of the Arsia Mons FSD is a thrust-block moraine. This moraine could have formed when proglacial or subglacial sediments were deformed by glacial thrusting, after stress built up due to a wet-based region of the glacier lying upslope of a cold-based region. The ribbed moraines located farther up ice (not shown) record a similar arrangement of local wet-based conditions. 207 Figure 13. Summary schematic showing the relationship of inferred glaciovolcanic and wet-based / polythermal glacial landforms in the Arsia Mons FSD (not to scale). Right: Enhanced ice flow and inferred polythermal conditions are recorded by ribbed moraines, streamlined knobs, and drop moraines that indicate a protruding lobe of ice centered on the subglacially erupted Northwest Plateau; fluvial channels that emerge beyond this suite of basal deposits are consistent with this assertion. Left: In the northeastern region of the FSD, the presence of numerous glaciovolcanic edifices upslope, ribbed moraine further downslope, and thrust-block moraine at the terminus of the deposit are all consistent with a former transition from local wet-based conditions upglacier to cold- based conditions downglacier. 208 Chapter 4 Remnant Buried Ice in the Equatorial Regions of Mars: Morphological Indicators Associated with the Arsia Mons Tropical Mountain Glacier Deposits Kathleen E. Scanlon,1 James W. Head,1 and David R. Marchant2 1 Department of Earth, Environmental and Planetary Sciences, Brown University 324 Brook Street, Box 1846, Providence, RI, 02912, USA 2 Department of Earth and Environment, Boston University 685 Commonwealth Avenue, Boston, MA, 02215, USA Published June 2015 in Planetary and Space Science, Volume 111, p. 144-154 doi: 10.1016/j.pss.2015.03.024. 209 Abstract The fan-shaped deposit (FSD) on the western and northwestern flanks of Arsia Mons is the remnant of tropical mountain glaciers, deposited several tens to hundreds of millions of years ago during periods of high spin-axis obliquity. Previous workers have argued that the Smooth Facies in the FSD contains a core of ancient glacial ice. Here, we find evidence that additional glacial ice remains preserved within several other landforms in the Smooth Facies and Ridged Facies. These include landforms that we interpret as kame and kettle topography on the basis of their distribution, size, and morphologies ranging progressively from knobs to degraded knobs to pits. We argue that some moraines in the Ridged Facies are ice-cored on the basis of their interactions with lava flows and the axial troughs at the crests of some moraines. We also argue that dunes with axial troughs, found in and surrounding the FSD, are the remnants of sediment-covered snow dunes formed by reworking of snow or glacial ice, and that the axial troughs form as tension cracks in the sediment and deepen by sublimation of the underlying ice. Long- term preservation of water ice in equatorial environments is assisted by a meters- to decameters-thick debris cover (lag) formed from sublimation of dirty ice, as well as burial beneath volcanic tephra and eolian deposits. This ancient ice has a high preservation potential for biosignatures, and could provide information on martian climate history and changes in the composition of the martian atmosphere over time, as well as resources for human exploration. 1. Introduction 210 The western and northwestern flanks of the equatorial Tharsis Montes volcanoes were covered by cold based mountain glaciers as recently as 125–220 million years ago (Kadish et al., 2014), as evidenced by the morphology, stratigraphic relationships, and spatial distribution of landforms in the fan-shaped deposits (FSDs) on each volcano (Williams, 1978; Lucchitta, 1981; Head and Marchant, 2003; Shean et al., 2005, 2007; Kadish et al, 2008a; Scanlon et al., 2014a,b). This geomorphologic evidence is bolstered by climate and glacial flow models that predict snow accumulation and ice flow in those regions during periods of high spin-axis obliquity (Forget et al., 2006; Fastook et al., 2008). Following a return to lower obliquity and the resulting change in climate conditions, the glacial ice ablated and returned to higher latitudes and the poles (Head et al.,2003; 2006a,b), leaving the Tharsis Montes fan-shaped deposits. A major question is whether buried ice still remains in some of these deposits, despite the peak insolation and relatively high temperatures expected at equatorial latitudes now and in the recent past (e.g. Mellon and Jakosky 1993, 1995; Mellon et al.,1997). Morphological evidence for buried present-day water ice in the tropics and mid- latitudes of Mars can be generally divided into three categories, namely: (1) Surface textures attributed to partial removal of ice. These include sublimation pits or hollows (e.g. Mustard et al., 2001; Mangold, 2003; Kadish et al., 2008b), scalloped depressions (e.g. Lefort et al., 2010; Séjourné et al.,2011), sublimation polygons and “brain terrain” (e.g. Levy et al., 2008; 2009), and other dissected terrains such as “basketball texture” (e.g. Head et al., 2003) or “ridge and valley texture” (Pierce and Crown, 2003; Chuang and Crown, 2005). 211 (2) Topographic profile. Lobate debris aprons (LDA), lineated valley fill (LVF), and concentric crater fill (CCF) on Mars have been interpreted as debris-covered glaciers with remnant ice cores, partially on the basis of the glacier-like convex upward topographic profiles at their margins (e.g. Mangold and Allemand, 2001; Holt et al., 2008; Head et al., 2010; Levy et al., 2010). (3) Unusual crater morphologies. “Ring-mold” craters (Kress and Head, 2008) have been interpreted as resulting from impacts into buried ice on the basis of their size- frequency distribution, which is consistent with smaller impacts not penetrating far enough to reach the buried ice; their annular moats, which are a characteristic feature of experimental impacts into ice-rich substrates; and the apparent degradation sequence represented by the range of ring-mold crater morphologies (Pedersen and Head, 2010). Pedestal craters, perched craters and excess ejecta craters (e.g. Kadish and Head, 2011) are interpreted to form by impacts into an ice-rich substrate, where either the impact process itself (in the case of pedestal craters) or the excavation of rocky material from underneath the ice-rich layer (in the case of perched and excess ejecta craters) creates a surface deposit that protects the ice-rich material immediately surrounding the crater against sublimation. Remnant ice at equatorial latitudes on Mars is of potential interest as an exploration target for several reasons. Gas bubbles preserved in terrestrial ancient ice can be used to develop time series for the molecular and isotopic composition of the atmosphere (e.g. Alley, 2000; Lüthi et al., 2008; Kobashi et al.,2011; Capron et al., 2012; Bazin et al., 2013; Rhodes et al., 2013). If a reliable chronology and isotopic baseline could be developed for Mars, then these data would be particularly useful, as they could 212 potentially help constrain orbital parameter variations prior to those that can be calculated a priori (Laskar et al., 2004). The Arsia Mons FSD has been suggested as a well-suited target for future human missions (e.g. Levine et al., 2010), and ice deposits within the FSD would offer a potential water and fuel resource for human exploration (e.g. Sridhar et al., 2004). At ~166,000 km2 in area, the Arsia Mons FSD (Head and Marchant, 2003; Shean et al., 2005, 2007; Scanlon et al., 2014a,b) is the largest of the Tharsis Montes FSDs (Figure 1). Crater counts indicate that the FSD has been in place for ~210 Ma (Kadish et al., 2014). The Smooth Facies, one of the geomorphologic units in the FSDs (Zimbelman and Edgett, 1992; Scott and Zimbelman, 1995), has been interpreted as remnant alpine- like debris-covered glaciers (Head and Marchant, 2003). Likewise, Shean et al. (2007) suggest that lineated debris displaying concentric ridges and partially filling tectonic graben higher up the volcanic edifice is also cored by glacier ice. The convex topography of these deposits (Shean et al., 2007), as well as the morphologic indicators of active flow (Head and Marchant, 2003; Marchant and Head, 2007) and the unique morphology of superimposed craters (Head and Weiss, 2014), suggest that buried ice 100 to 300 m thick may still be present at depth. In this contribution, we expand the search for buried ice and review several other classes of landforms in the FSD that have not been previously described, and whose morphology indicates that remnant ice may still be present. 2. Data and Methods Images in this study are from the Mars Reconnaissance Orbiter (MRO) Context Camera (CTX), with ~5 meter per pixel resolution (Malin et al., 2007), augmented with 213 images from the High Resolution Stereo Camera (HRSC) at 10–30 meters per pixel resolution (Neukum and Jaumann, 2004). Topographic data is from the Mars Orbital Laser Altimeter (MOLA) at ~463 meters per pixel resolution (Zuber et al., 1992; Smith et al., 1999) and, where available, HRSC-derived Digital Elevation Maps (DEMs) with ~100 meters per pixel resolution (Dumke et al., 2008). Contour maps were created using the Spatial Analyst toolkit in ArcMap 10.0. 3. Landforms Interpreted to be Indicative of Remnant Ice On Earth, remnant patches of buried glacier ice may occur wherever overlying debris is sufficiently thick to retard ice ablation. Examples include ice-cored moraines, detached blocks of ice buried beneath proglacial sediment, and remnant, stagnant ice buried beneath thick sublimation till (e.g. Hambrey, 1984; Marchant et al., 2002; Evans, 2009; Swanger et al., 2010; Irvine-Flynn et al., 2011; Lacelle et al., 2011; Monnier et al., 2012). In the coldest and driest region of the Mars-like Antarctic Dry Valleys, 40Ar/39Ar ages of volcanic ash deposits indicate that underlying remnant glacier ice has been preserved for millions of years (Sugden et al., 1995; Marchant et al., 2002; Kowalewski et al., 2006, 2012). We propose that the morphology of several landforms adjacent to Arsia Mons suggests that ice millions of years old is also present in the Arsia Mons FSD. 3.1 Pit-and-knob terrain. Near the northern edge of the FSD is a field of mounds (“knobs”) and shallow topographic depressions (“pits”; Figures 1 and 2). Each knob is up to 1 km in diameter. 214 Interspersed among the knobs are pits of similar size and shape to the knobs (Figures 2, 3). The pits and knobs are generally aligned, and the lines on which they fall are concentric with the outline of the Smooth Facies (Figures 1, 2) and with drop moraines left by a relatively young debris-covered glacier extending from a nearby graben (Shean et al., 2007). Many of the smaller knobs are surrounded by shallow annular depressions (“moats”), and some pits have what appear to be degraded knobs at their centers (Figure 3). We propose that the pit-and-knob terrain is ice-cored and that the landforms represent a progression in which gradual loss of ice via sublimation causes topographic inversion, with knobs becoming moated knobs, then pits with degraded knobs, and finally pits (Figure 4). In terrestrial zones of rapid ice retreat, blocks of ice detached from the retreating edge of a glacier may become partially buried beneath glacial outwash (Thwaites, 1926; Price, 1969; Fay, 2002; Russell et al., 2010; Evans, 2011; and Knight, 2012). When the blocks eventually melt, they leave “kettle holes” where the blocks formerly stood. Fields of pits interpreted as kettle holes have been observed on Mars in the circumpolar Dorsa Argentea Formation (Dickson and Head, 2006). Together, (1) the similarity in size and shape between the pits and knobs, (2) the genetic relationship implied by the presence of knobs with moats, and (3) the co-alignment of the pits and knobs with the outline of the Smooth Facies and with lineations in the Smooth Facies (Figure 5) suggest that the Arsia FSD pit-and-knob terrain may have resulted from backwasting of ice in the Smooth Facies and subsequent burial of the isolated ice blocks, analogous to the formation of terrestrial kettled outwash plains (Figure 6). Because of the cold Amazonian climate, however, the ice blocks would have sublimed rather than melted to leave the pits behind, 215 and the sediment that embayed them would have been volcanic tephra, englacial debris, or aeolian sediment, rather than glacial outwash as in terrestrial kettled plains. The concentric fractures surrounding many of the pits (Figure 3) suggest that near-surface sediment, possibly cemented by pore ice, moved downslope toward pit centers as underlying ice sublimed; similar patterns can be observed on Earth where the removal of blocks of buried ice causes concentric fractures to form in the overlying sediment (Sanford, 1959; Dickson and Head, 2006). This sediment cover could have armored some of the ice blocks against further sublimation, leaving the present-day knobs. Alternatively, the pit-and-knob terrain may have formed in a manner analogous to terrestrial “controlled moraine” (e.g. Evans, 2009; Szuman and Kasprzak, 2010; Bennett and Evans, 2012; Lakeman and England, 2012). When variable concentrations of debris within or on top of glaciers (e.g. Mackay et al., 2014) are shaped into belts by glacial flow, bands of debris can isolate patches of stagnant ice as the glacier retreats. Incomplete loss of this dead ice results in linear arrangements of mounds and kettle holes, with the continuity and linearity of the mounds breaking down as the removal of ice continues to completion. As with the “backwasting” formation model, the knob-to-pit progression (Figure 4), the linearity of the remaining knobs (Figures 1, 2, 5), and the concentric fractures surrounding pits (Figure 3) suggest that these features had ice cores at formation and that the removal of ice from these features has not proceeded to completion. This “controlled moraine” model (Figure 7) accounts for the linear arrangement of the pits and knobs without requiring widespread accumulation of ice blocks to have occurred at the ice margins. We therefore favor this model of pit-and-knob terrain development. 216 3.2 Drop moraines with linear troughs. At the northwestern edge of the deposit, some of the drop moraines (Head and Marchant, 2003) in the Ridged Facies have linear troughs along their crests (Figure 8). These moraines are typically ~100 m wide, and some are continuous for over 100 km. Other nearby moraines have similar dimensions but lack the characteristic crest troughs. We interpret these moraines to have developed their trough morphology by the loss of ice via sublimation. In terrestrial settings, ice-cored moraines can develop at the margins of glaciers when bands of debris within a glacier isolate small masses of ice from the main body of ice as the glacier retreats (e.g. Evans, 2009). The hypothesis that moraines in the FSD may contain remnant ice is also supported by the lone subaerial lava flow in the Ridged Facies, which has a chaotic texture interpreted to have resulted from interactions between the lava flows and ice in and around the moraines at the time of flow emplacement (Scanlon et al., 2014a). If ice was present in the moraines when lava flowed over them, the lava would be expected to interact with it by melting and collapse (e.g. Edwards et al., 2012), as well as explosively (e.g. Belousov et al., 2011), imparting a chaotic texture to the flows. 4. Postglacial Ice-Related Landforms Dispersed throughout the northwestern edge and surroundings of the Arsia FSD are relatively short (~2 km, but some as much as 9 km) elongate ridges with axial troughs (Figure 11). The width of the ridges in any one population of ridges is highly uniform, but typical widths vary between regions from ~100 m to ~350 m. Their distribution is not restricted to any underlying unit of the FSD, and they often appear superposed on and 217 draping the contacts and features of the underlying FSD units (Figure 10). In contrast to the characteristics of the concentric drop moraines of the Ridged Facies, they are consistently oriented southwest-to-northeast wherever they occur, and they are straight rather than curved in plan view. They are also continuous over much shorter distances than the drop moraines with linear troughs. They are often concentrated in local topographic lows rather than being uniformly located within a particular unit. In general, they appear dune-like in nature, but are distinguished from typical dune forms in having a trough along their long axis. On the basis of the physical proximity of these features to the Arsia Mons glacier deposits, and the likelihood that at least some snow and ice is deposited in the Tharsis FSD regions whenever spin-axis obliquities measurably increase from current values (Forget et al., 2006; Schon et al., 2012), we suggest that this morphology results from the sublimation of ice along the crests of ice-rich dunes. Pedestal craters found throughout the Tharsis region and dated to 12–13 Ma (Schon et al., 2012) suggest that several meters of ice covered the Tharsis region in a recent phase of moderately high obliquity. Due to the superposition of the dunes upon the other units of the Arsia Mons FSD, we propose that deposits such as this later ice cover may have been the source of the ice that formed the dunes displaying linear troughs. The greater density of dunes within the FSD (Figure 10) suggests that the dunes may also have been built by reworking of the ice and debris from the FSD itself. The morphology of these ridges suggests the removal of a volatile component. There are two possibilities for the nature of the dust-ice mixture that could give rise to the troughs at the crest of the ridges. First (Figure 11a), the ridges could have formed as 218 mixed snow-ice dunes formed during snow deposition at higher obliquity. When the spin- axis obliquity lowered such that ice was no longer stable at the equator (e.g. Jakosky and Carr, 1985; Mellon and Jakosky, 1993), the snow component would have sublimed and the dust component would have become concentrated in the outermost layers (Figure 11b), analogous to the development process of the martian Latitude-Dependent Mantle and some types of terrestrial loess (Mustard et al., 2001; Head et al., 2003). If the sides of the dunes were sufficiently steep, these dry outer layers would have slumped down the ridge sides, forming debris-rich piles at either side of the crest and exposing more ice to sublimation (Figure 11c). A similar process helps drive topographic inversion cycles in ice-cored moraines and creates ring-shaped “circular moraine features” from sufficiently tall debris-covered dead ice blocks on Earth (Ebert and Kleman, 2004). In concert with this mechanism, continued wind shear could cause dust to be removed from the crest, exposing buried ice-rich material to preferential sublimation. Second, the ridges could have formed as snow dunes that were later covered by a layer of dust or tephra (Figure 12a,b). Because the least compressive stress in a topographic ridge is oriented horizontally away from the axis of the ridge (Fiske and Jackson, 1972; McTigue and Mei, 1981; Dieterich, 1988; Rubin and Rubin, 2013), long tensile cracks aligned with the dune crests could be expected to form in the sediment cover (Figure 12c). This tendency could be further enhanced if the surface sediment layer was not frozen to the underlying snow and ice, in which case it could slump to either side of the ridge crest. Fractures are observed along the crests of debris-covered snow ripples in the Antarctic Dry Valleys (compare Figures 9 and 13) and develop parallel to topographic contours on niveo-aeolian dunes and beds in Alaska as their snow 219 component melts (Koster and Dijkmans, 1988). These fractures in the protective dust cover would enhance sublimation directly beneath them by exposing the underlying snow (Mangold, 2003; 2011), eventually leaving a hollow along the ridge axis (Figure 12d,e). On the basis of the similarity of these features to terrestrial debris-covered snow dunes, and the fact that plausible heights for the ripples are lower than the ice block heights that form ring-shaped circular moraines on Earth (and are thus more likely to create a single ridge of debris than two parallel ridges; Ebert and Kleman, 2004), we currently favor this latter interpretation. High-resolution topographic data will help distinguish between these mechanisms more conclusively by constraining the height and side slopes of individual ridges. Unlike the drop moraines described in section 3, there are no dunes in the FSD that are similar in shape, size, and distribution to the crest-trough dunes but which lack the troughs. This suggests that ice removal in the dunes may have proceeded to completion. By analogy with the Antarctic debris-covered snow ripples, ice may still be present beneath the dunes, but this cannot be determined from image data alone. The primary importance of these landforms is therefore not as a likely reservoir of present- day ice, but rather as an additional indicator of the extent and aeolian transport of equatorial ice in a recent high-obliquity excursion. 5. Discussion How much ice remains within the FSDs, what data aside from geomorphologic observations indicate the presence of this ice, and where does this remnant ice fit in the timeline of Amazonian climate change? For comparison, the best-studied deposits of non- 220 polar Amazonian ice are the lobate debris aprons (LDAs), which are mid- to late- Amazonian aged debris-covered remnant glaciers found throughout the mid-latitudes of Mars (e.g. Head et al., 2010). Radar data is consistent with the hypothesis that LDAs are composed of massive water ice covered by a layer of debris 0.5–10 m thick (Holt et al., 2008; Plaut et al., 2009). If the age of the FSD and the LDAs are similar, and the temperature and humidity of the environments surrounding them (and hence the stability of ice in those environments) are also similar, then the depth to massive ice would be expected to be somewhat greater for landforms in the equatorial FSDs than for the mid- latitude LDAs (e.g. Schorghofer and Forget, 2012). Many of our proposed ice-cored features at Arsia Mons are several times greater than 10 m in height, such that a debris cover thick enough to preserve an ice core could remain. The fact that the characteristic crest troughs are most evident in the relatively small moraines and dunes may in fact be due to their smaller size; the debris cover on larger features may be sufficient that no significant volume of ice has yet been removed, or that the thick surface debris masks topography associated with underlying ice loss. The elevated concentration of hydrogen on the western sides of Arsia and Pavonis Mons relative to their eastern sides (as indicated by local minima in epithermal neutron fluxes; Boynton et al., 2002; Elphic et al., 2005) is also consistent with the hypothesis that some ice remains in the FSDs. Elphic et al. (2005), making the simplifying assumption that the hypothesized ice-rich deposits are pure ice overlain by dry sublimation till in order to obtain a first order estimate, calculated that ice could lie ~60 cm below the surface. Since the widespread dunes in the FSD suggest recent or ongoing aeolian activity, it is possible that wind has removed debris from the deposit such that the 221 thickness of debris currently overlying any ice in the deposit is less than the amount that was required to preserve it to the present day. Radar data could potentially bolster the geomorphological evidence for remnant ice in the FSDs, but have not been clear. For example, it has also been suggested that the Pavonis Mons FSD contains remnant ice in its Smooth Facies deposits (Shean et al., 2005). Recent SHARAD radar profiles, however, did not detect strong reflections in this unit as they did for the LDAs, and as would be expected for an internally layered ice core (Campbell et al., 2013). Head and Weiss (2014) presented geomorphologic evidence for up to several hundred meters of present-day ice under a ≥ 16 m thick debris cover in the Smooth Facies at Pavonis and Arsia Mons. They suggest that the lack of strong radar reflections could be explained if any ice in the Smooth Facies is intermingled with volcanic tephra. Due to the proximity of the Arsia Mons volcano, tephra would be expected to be more abundant in the FSD than in the LDAs. The total amount of ice potentially remaining in the landforms described in this paper cannot be estimated at present due to the absence of topographic data that can resolve the smaller classes of potentially ice-cored landforms, e.g. Ridged Facies moraines. The scale of the kettle pits and hypothesized ice-cored knobs (section 3.1) is larger, however, and these landforms are resolved in HRSC DEMs (Figure 14). Within the area of HRSC DEM coverage, knobs stand ~15–80 m high, whereas pits are typically 5–10 m deep. If this height difference is entirely caused by the removal of volatiles, and if these dimensions are typical for the pits and knobs not covered by HRSC DEMs, the hundreds of knobs remaining in the FSD could each contain a body of ice 20–90 m thick at their cores. 222 6. Conclusions The geomorphology of three classes of landform in the Arsia Mons FSD suggests that the deposit contains more remnant ice than previously thought. Evidence for extant ice in the bulk of the Smooth Facies has been described by previous researchers (Shean et al., 2007; Head and Weiss, 2014); the evidence we present for extant ice in some Ridged Facies drop moraines and in small landforms at the margins of the Smooth Facies increases the total volume of proposed ice in the FSD. Trough morphologies suggest that ice has been removed from some but not all of the glacial moraines in the deposit, and the chaotic surface texture of volcanic flows to the northwest of the deposit suggests that they interacted with ice-cored moraines. The size and morphology of pits and knobs near the northern edge of the FSD suggest that the knobs contain ice that is armored by debris. These deposits should be added to the volumes of sequestered ice mapped throughout the mid-latitudes (Levy et al., 2014). Fields of dunes with linear troughs along their crests suggest that windblown snow was widespread across the region in the Amazonian. The knobs and moraines examined in this study represent a potential reservoir of buried, present-day equatorial ice, in addition to the ice previously estimated (Shean et al., 2007; Head and Weiss, 2014) to remain in the Smooth Facies. These features are present in several regions of the deposit, near other landforms of climatological, volcanological, and possible astrobiological interest (Scanlon et al., 2014a,b). Because of its location near the equator, the Arsia Mons FSD may be a good target for human missions to study martian ice while avoiding the logistical difficulties of a polar mission (e.g. Cockell, 2001). 223 Acknowledgements We gratefully acknowledge support from the NASA Graduate Student Researchers Program (Grant NNX12AI39H) to KES, from the Mars Data Analysis Program (Grant NNX11AI81G) and the Mars Express High-Resolution Stereo Camera (HRSC) Investigation Team (JPL 1488322) to JWH, and from NSF Polar Programs (Grant ANT-0944702) to DRM. We thank David Hollibaugh Baker for the use of his photographs in Figure 13, and an anonymous reviewer for their constructive suggestions. References Alley, R. B. (2000), Ice-core evidence of abrupt climate changes, Proceedings of the National Academy of Sciences, 97(4), 1331–1334. Bazin, L., et al. (2013), An optimized multi-proxy, multi-site Antarctic ice and gas orbital chronology (AICC2012): 120–800 ka, Climate of the Past, 9, 1715–1731. Belousov, A., B. Behncke, and M. Belousova (2011), Generation of pyroclastic flows by explosive interaction of lava flows with ice/water-saturated substrate, Journal of Volcanology and Geothermal Research, 202, 60–72. Bennett, G. L., and D. J. A. Evans (2012), Glacier retreat and landform production on an overdeepened glacier foreland: the debris-charged glacial landsystem at Kvíárjökull, Iceland, Earth Surface Processes and Landforms, 37, 1584–1602. Boynton, W. V., et al. (2002), Distribution of hydrogen in the near surface of Mars: evidence for subsurface ice deposits, Science, 297(5578), 81-85. 224 Campbell, B. A., N. E. Putzig, L. M. Carter, G. A. Morgan, R. J. Phillips, and J. J. Plaut (2013), Roughness and near-surface density of Mars from SHARAD radar echoes, Journal of Geophysical Research: Planets, 118, 436–450. Capron, E., A. Landais, J. Chappellaz, D. Buiron, H. Fischer, S. J. Johnsen, J. Jouzel, M. Leuenberger, V. Masson-Delmotte, and T. F. Stocker (2012), A global picture of the first abrupt climatic event occurring during the last glacial inception, Geophysical Research Letters, 39, L15703. Chuang, F. C., and D. A. Crown (2005), Surface characteristics and degradational history of debris aprons in the Tempe Terra/Mareotis fossae region of Mars, Icarus, 179, 24–42. Cockell, C. S. (2001), Martian polar expeditions: problems and solutions, Acta Astronautica, 49(12), 693-706. Dickson, J. L., and J. W. Head (2006), Evidence for an Hesperian-aged south circum- polar lake margin environment on Mars, Planetary and Space Science, 54, 251- 272. Dumke, A., M. Spiegel, R. Schmidt, and G. Neukum (2008), High-resolution digital terrain models and ortho-image mosaics of Mars: Generation on the basis of Mars-Express HRSC data, presented at Lunar and Planetary Science Conference, XXXIX, abstract #1910. Ebert, K., and J. Kleman (2004), Circular moraine features on the Varanger Peninsula, northern Norway, and their possible relation to polythermal ice sheet coverage, Geomorphology, 62(3-4), 159–168. 225 Edwards, B., E. Magnússon, T. Thordarson, M. T. Guđmundsson, A. Höskuldsson, B. Oddsson, and J. Haklar (2012), Interactions between lava and snow/ice during the 2010 Fimmvörðuháls eruption, south-central Iceland, Journal of Geophysical Research: Solid Earth, 117, B04302. Elphic, R. C., W. C. Feldman, T. H. Prettyman, R. L. Tokar, D. J. Lawrence, J. W. Head, and S. Maurice (2005), Mars Odyssey Neutron Spectrometer water-equivalent hydrogen: Comparison with glacial landforms on Tharsis, presented at Lunar and Planetary Science Conference XXXVI, abstract #1805. Evans, D. J. A. (2009), Controlled moraines: origins, characteristics and palaeoglaciological implications, Quaternary Science Reviews, 28(3-4), 183–208. Evans, D. J. A. (2011), Glacial landsystems of Satujökull, Iceland: A modern analogue for glacial landsystem overprinting by mountain icecaps, Geomorphology, 129, 225–237. Fastook, J. L., J. W. Head, D. R. Marchant, and F. Forget (2008), Tropical mountain glaciers on Mars: Altitude-dependence of ice accumulation, accumulation conditions, formation times, glacier dynamics, and implications for planetary spin-axis/orbital history, Icarus, 198(2), 305-317. Fay, H. (2002), Formation of kettle holes following a glacial outburst flood (jökulhlaup), Skeidarársandur, southern Iceland, in The Extremes of the Extremes: Extraordinary Floods, eds. A. Snorrason, H. P. Finnsdóttir, and M. E. Moss, 205- 210. International Association of Hydrological Sciences. 226 Forget, F., R. Haberle, F. Montmessin, B. Levrard, and J. Head (2006), Formation of glaciers on Mars by atmospheric precipitation at high obliquity, Science, 311(5759), 368-371. Hambrey, M. J. (1984), Sedimentary processes and buried ice phenomena in the pro- glacial areas of Spitsbergen glaciers, Journal of Glaciology, 30(104), 116-119. Head, J. W., and D. R. Marchant (2003), Cold-based mountain glaciers on Mars: Western Arsia Mons, Geology, 31(7), 641-644. Head, J.W., J. F. Mustard, M. A. Kreslavsky, R. E. Milliken, and D. R. Marchant (2003), Recent ice ages on Mars, Nature, 426, 797-802. Head, J. W., D. R. Marchant, M. C. Agnew, C. I. Fassett, and M. A. Kreslavsky (2006a), Extensive valley glacier deposits in the northern mid-latitudes of Mars: Evidence for Late Amazonian obliquity-driven climate change, Earth and Planetary Science Letters, 241, 663-671. Head, J. W., A. L. Nahm, D. R. Marchant, and G. Neukum (2006b), Modification of the dichotomy boundary on Mars by Amazonian mid-latitude regional glaciation, Geophysical Research Letters, 33, L08S03. Head, J. W., D. R. Marchant, J. L. Dickson, A. M. Kress, and D. M. H. Baker (2010), Northern mid-latitude glaciation in the Late Amazonian period of Mars: Criteria for the recognition of debris-covered glacier and valley glacier landsystem deposits, Earth and Planetary Science Letters, 294, 306-320. Head, J. W., and D. K. Weiss (2014), Tropical mountain glacier deposits on Mars: Preservation of ancient ice at Pavonis and Arsia Mons, submitted to Icarus. 227 Holt, J. W., A. Safaeinili, J. J. Plaut, J. W. Head, R. J. Phillips, R. Seu, S. D. Kempf, P. Choudhary, D. A. Young, N. E. Putzig, D. Biccari, and Y. Gim (2008), Radar sounding evidence for buried glaciers in the southern mid-latitudes of Mars, Science, 322, 1235-1238. Jakosky, B. M., and M. H. Carr (1985), Possible precipitation of ice at low latitudes of Mars during periods of high obliquity, Nature, 315(6020), 559-561. Kadish, S. J., J. W. Head, R. L. Parsons, and D. R. Marchant (2008a), The Ascraeus Mons fan-shaped deposit: Volcano-ice interactions and the climatic implications of cold-based tropical mountain glaciation, Icarus, 197(1), 84-109. Kadish, S. J., J. W. Head, N. G. Barlow, and D. R. Marchant (2008b), Martian pedestal craters: Marginal sublimation pits implicate a climate-related formation mechanism, Geophysical Research Letters, 35, L16104. Kadish, S. J., and J. W. Head (2011), Impacts into non-polar ice-rich paleodeposits on Mars: Excess ejecta craters, perched craters and pedestal craters as clues to Amazonian climate history, Icarus, 215, 34-46. Kadish, S. J., J. W. Head III, J. L. Fastook, and D. R. Marchant (2014), Middle to Late Amazonian tropical mountain glaciers on Mars: The ages of the Tharsis Montes fan-shaped deposits, Planetary and Space Science, 91, 52-59. Knight, J. (2012), Glaciotectonic sedimentary and hydraulic processes at an oscillating ice margin, Proceedings of the Geologists’ Association, 123, 714–727. Kobashi, T., K. Kawamura, J. P. Severinghaus, J.-M. Barnola, T. Nakaegawa, B. M. Vinther, S. J. Johnsen and J. E. Box (2011), High variability of Greenland surface 228 temperature over the past 4000 years estimated from trapped air in an ice core, Geophysical Research Letters, 38, L21501. Kowalewski, D.E., Marchant, D.R., Head, J.W., and Jackson, D.W., 2012. A 2D model for characterizing first-order variablility in sublimation of buried glacier ice, Antarctica: Assessing the influence of polygon troughs, desert pavements, and shallow-subsurface salts. Permafrost and Periglacial Processes 23, 1-14. DOI: 10.1002/ppp.731. Kowalewski, D. E., Marchant, D.R., Levy, J.S., and Head, J.W. III. 2006. Quantifying low rates of summertime sublimation for buried glacier ice in Beacon Valley, Antarctica. Antarctic Science 18, 421-428. Kress, A. M., and J. W. Head (2008), Ring-mold craters in lineated valley fill and lobate debris aprons on Mars: Evidence for subsurface glacial ice, Geophysical Research Letters, 35, L23206. Lakeman, T. R., and J. H. England (2012), Paleoglaciological insights from the age and morphology of the Jesse moraine belt, western Canadian Arctic, Quaternary Science Reviews, 47, 82–100. Laskar, J., A.C.M. Correia, M. Gastineau, F. Joutel, B. Levrard, and P. Robutel (2004), Long term evolution and chaotic diffusion of the insolation quantities of Mars, Icarus, 170(2), 343–364. Lefort, A., P.S. Russell, and N. Thomas (2010), Scalloped terrains in the Peneus and Amphitrites Paterae region of Mars as observed by HiRISE, Icarus, 205, 259– 268. 229 Levine, J. S., J. B. Garvin, and J. W. Head (2010), Martian geology investigations: Planning for the scientific exploration of mars by humans, Part 2, Journal of Cosmology, 12, 3636-3646. Levy, J., J. W. Head, D. R. Marchant, and D. Kowalewski (2008), Identification of sublimation-type thermal contraction crack polygons at the proposed NASA Phoenix landing site: Implications for substrate properties and climate-driven morphological evolution, Geophysical Research Letters, 35, L04202. Levy, J. S., J. W. Head, and D. R. Marchant (2009), Concentric crater fill in Utopia Planitia: History and interaction between glacial "brain terrain" and periglacial mantle processes, Icarus, 202, 462-476. Levy, J. S., J. W. Head III, and D. R. Marchant (2010), Concentric crater fill in the northern mid-latitudes of Mars: Formation processes and relationships to similar landforms of glacial origin, Icarus, 209, 390-404. Levy, J. S., C. I. Fassett, J. W. Head, C. Schwartz, and J. L. Watters (2014), Sequestered glacial ice contribution to the global martian water budget: Geometric constraints on the volume of remnant, mid-latitude debris covered glaciers, Journal of Geophysical Research: Planets, in press. Lucchitta, B. K. (1981), Mars and Earth - Comparison of cold-climate features, Icarus, 45(2), 264-303. Lüthi, D., et al. (2008), High-resolution carbon dioxide concentration record 650,000– 800,000 years before present, Nature, 453, 379-382. Mackay, S.L., D. R. Marchant, J. W. Head, and J. L. Lamp (2014), Cold-based debris- covered glaciers: Evaluating their potential as climate archives through studies of 230 ground-penetrating radar and surface morphology, Journal of Geophysical Research,Earth Surface, 119, doi:10.1002/ 2014JF003178. Malin, M. C., et al. (2007), Context Camera Investigation on board the Mars Reconnaissance Orbiter, Journal of Geophysical Research: Planets, 112(E05). Mangold, N., and P. Allemand (2001), Topographic analysis of features related to ice on Mars, Geophysical Research Letters, 28, 407–410. Mangold, N. (2003), Geomorphic analysis of lobate debris aprons on Mars at Mars Orbiter Camera scale: Evidence for ice sublimation initiated by fractures, Journal of Geophysical Research: Planets, 108(E4), 2-1-2-13. Mangold, N. (2011), Ice sublimation as a geomorphic process: A planetary perspective, Geomorphology, 126(1), 1-17. Marchant, D.R., Lewis, A., Phillips, W.C., Moore, E.J., Souchez, R., and Landis, G. P. 2002. Formation of patterned-ground and sublimation till over Miocene glacier ice in Beacon Valley, Antarctica. Geological Society of America Bulletin 114, 718-730. Marchant, D.R. and Head, J.W. 2007. Antarctic Dry Valleys: Microclimate zonation, variable geomorphic processes, and implications for assessing climate change on Mars. Icarus, v. 192, no. 1, p. 187-222, doi:10.1016/j.icarus.2007.06.018. Mellon, M. T., and B. M. Jakosky (1993), Geographic variations in the thermal and diffusive stability of ground ice on Mars, Journal of Geophysical Research: Planets, 98(E2), 3345–3364. Mellon, M. T., and B. M. Jakosky (1995), The distribution and behavior of Martian ground ice during past and present epochs, Journal of Geophysical Research: 231 Planets, 100(E6), 11781–11799. Mellon, M. T., B. M. Jakosky, and S. E. Postawko (1997), The persistence of equatorial ground ice on Mars, Journal of Geophysical Research: Planets, 102(E8), 19357– 19369. Mustard, J. F., C. D. Cooper, and M. K. Rifkin (2001), Evidence for recent climate change on Mars from the identification of youthful near-surface ground ice, Nature, 412(6845), 411-414. Neukum, G., and R. Jaumann (2004), HRSC: The High Resolution Stereo Camera of Mars Express, in Mars Express: The Scientific Payload, ed. A. Wilson, 17-35. European Space Agency. Pedersen, G. B., and J. W. Head (2010), Evidence of widespread degraded Amazonian- aged ice-rich deposits in the transition between Elysium Rise and Utopia Planitia, Mars: Guidelines for the recognition of degraded ice-rich materials, Planetary and Space Science, 58, 1953-1970. Pierce, T. L., and D. A. Crown (2003), Morphologic and topographic analyses of debris aprons in the eastern Hellas region, Mars, Icarus, 163, 46–65. Plaut, J. J., A. Safaeinili, J. W. Holt, R. J. Phillips, J. W. Head, R. Seu, N. E. Putzig, and A. Frigeri (2009), Radar evidence for ice in lobate debris aprons in the mid- northern latitudes of Mars, Geophysical Research Letters, 36, L02203. Price, R. J. (1969), Moraines, Sandar, Kames and Eskers near Breidamerkurjökull, Iceland, Transactions of the Institute of British Geographers, 46, 17-43. Rhodes, R. H., X. Faïn, C. Stowasser, T. Blunier, J. Chappellaz, J. R. McConnell, D. Romanini, L. E. Mitchell, and E. J. Brook (2013), Continuous methane 232 measurements from a late Holocene Greenland ice core: Atmospheric and in-situ signals, Earth and Planetary Science Letters, 368, 9–19. Russell, A. J., F. S. Tweed, M. J. Roberts, T. D. Harris, M. T. Gudmundsson, Ó. Knudsen, and P. M. Marren (2010), An unusual jökulhlaup resulting from subglacial volcanism, Sólheimajökull, Iceland, Quaternary Science Reviews, 29, 1363–1381. Sanford, R.A., (1959), Analytical and experimental study of simple geologic structures, Geological Society of America Bulletin, 70, 19–51. Scott, D., and J. Zimbelman (1995), Geologic map of Arsia Mons Volcano, Mars, US Geological Survey Miscellaneous Geologic Investigation Map I-2480. Scanlon, K. E., J. W. Head, L. Wilson, and D. R. Marchant (2014a), Volcano–ice interactions in the Arsia Mons tropical mountain glacier deposits, Icarus, 237, 315–339. Scanlon, K. E., J. W. Head, and D. R. Marchant (2014b), Local, volcanism-induced wet- based glacial conditions recorded in the Late Amazonian Arsia Mons tropical mountain glacier deposits, Icarus, in press. Schon, S. C., and J. W. Head III (2012), Decameter-scale pedestal craters in the tropics of Mars: Evidence for the recent presence of very young regional ice deposits in Tharsis, Earth and Planetary Science Letters, 317-318, 68-75. Schorghofer, N., and F. Forget (2012), History and anatomy of subsurface ice on Mars, Icarus, 220, 1112–1120. 233 Séjourné, A., F. Costard, J. Gargani, R.J. Soare, A. Fedorov, and C. Marmo (2011), Scalloped depressions and small-sized polygons in western Utopia Planitia, Mars: A new formation hypothesis, Planetary and Space Science, 59, 412–422. Shean, D. E., J. W. Head, and D. R. Marchant (2005), Origin and evolution of a cold- based tropical mountain glacier on Mars: The Pavonis Mons fan-shaped deposit, Journal of Geophysical Research: Planets, 110(E5). Shean, D. E., J. W. Head, J. L. Fastook, and D. R. Marchant (2007), Recent glaciation at high elevations on Arsia Mons, Mars: Implications for the formation and evolution of large tropical mountain glaciers, Journal of Geophysical Research: Planets, 112(E3). Smith, D. E., et al. (1999), The Global Topography of Mars and Implications for Surface Evolution, Science, 284(5419), 1495-1503. Sridhar, K. R., C. S. Iacomini, and J. E. Finn (2004), Combined H2O/CO2 solid oxide electrolysis for Mars in situ resource utilization, Journal of Propulsion and Power, 20(5), 892–901. Sugden, D. E., D. R. Marchant, N. Potter Jr, R. A. Souchez, G. H. Denton, C. C. Swisher III, and J.-L. Tison (2002), Preservation of Miocene glacier ice in East Antarctica, Nature, 376, 412-414. Swanger, K.M., Marchant, D.R., Kowalewski, D.E., and Head, J.W. 2010. Viscous flow lobes in central Taylor Valley, Antarctica: Origin as remnant buried glacial ice. Geomorphology, v. 120, p. 174-185. doi:10.1016/j.geomorph.2010.03.024. 234 Szuman, I., and L. Kasprzak (2010), Glacier ice structures influence on moraine development (Hørbye Glacier, Central Spitsbergen), Quaestiones Geographicae, 29, 65–73. Thwaites, F. T. (1926), The origin and significance of pitted outwash, The Journal of Geology, 34(4), 308-319. Williams, R. (1978), Geomorphic processes in Iceland and on Mars: A comparative appraisal from orbital images, Geological Society of America Abstracts with Programs, 10, 517. Zimbelman, J. R., and K. S. Edgett (1992), The Tharsis Montes, Mars - Comparison of volcanic and modified landforms, presented at Lunar and Planetary Science Conference, XXII, p. 31-44. Zuber, M. T., D. E. Smith, S. C. Solomon, D. O. Muhleman, J. W. Head, J. B. Garvin, J. B. Abshire, and J. L. Bufton (1992), The Mars-Observer Laser Altimeter Investigation, Journal of Geophysical Research: Planets, 97(E5). 235 Figure 1. Geomorphological unit map of the Arsia Mons fan-shaped deposit (FSD), after Zimbelman and Edgett (1992) and Scott and Zimbelman (1995); reproduced from Scanlon et al. (2014) and annotated. Red lines denote large volcanic graben, white lines denote contacts between units, and black lines denote the outlines of glaciovolcanic landforms. Closed and open green circles in the smooth facies denote pits and knobs, respectively. The regions where moraines and dunes with linear troughs (Sections 3.2 and 4, respectively) are located are marked by shaded black and white ellipses, respectively. The area shown in Figure 2 is denoted by a white box. THEMIS 100 m / pixel daytime image mosaic. This and all other images in this paper are oriented with north toward the top of the image. 236 Figure 2. Pit (white arrow) and knob (black arrow) terrain in the Arsia Mons fan-shaped deposit (see Figure 1 for location). Pits and knobs form lines concentric to the border of the Smooth Facies. The area shown in Figure 3 is denoted by a white box. Some knobs appear to be surrounded by moats (red arrow), implying a progression from knob to pit (see also Figure 4). CTX image P08_004056_1741_XI_05S125W. 237 Figure 3. Pit-and-knob terrain: (a) Closeup view of pits and knobs outlined by the white box in Figure 2. CTX image P08_004056_1741_XI_05S125W. (b) Sketch map of area in Figure 3a; pits are shown in red, knobs in yellow, and fractures in blue. a b 238 Figure 4. Examples of knobs (a), moated knobs (b), pits with remnant knobs (c), and pits (d), from CTX images P08_004056_1741_XI_05S125W and P16_007392_1743_XN_05S126W. a b 239 c d 240 Figure 5. Pits and knobs fall along lines concentric to the boundaries of the Smooth Facies. (a) THEMIS image mosaic. (b) Sketch map; the Smooth Facies is shown in orange, the Knobby Facies in blue, pits as filled green circles, and knobs as open green circles. a b 241 Figure 6. Comparison of kettle hole formation process on Earth (right) and a “backwasting” model of pit and knob terrain formation in the Arsia Mons fan-shaped deposit (left). As a glacier (a) recedes, blocks of ice separate from its terminus (b). In a warm-based, terrestrial glacier, outwash from the receding glacier will partially bury (or in some cases, not shown, completely bury) these blocks; on Amazonian Mars, atmospheric dust or volcanic tephra covers the ice blocks before they fully sublime (c). When the ice fully melts or sublimes, shallow pits are left behind in the sediment; in the case of Mars, some ice has been armored by debris fall and remains as debris-covered knobs (d). a b c d 242 Figure 7. A “controlled moraine” model of pit-and-knob terrain formation in the Arsia Mons fan-shaped deposit. (a) Glacial flow concentrates debris into discrete bands within a debris-covered glacier (or thickens bands of debris atop the glacier, resulting in similar variable preservation of the underlying ice). (b) As downwasting occurs, the bands of debris isolate masses of ice, concentric to the outlines of the glacier. (c) As ice removal proceeds, the ice-cored ridges become less continuous and form knobs or elongate mounds. Complete removal of ice in some mounds (by sublimation) results in the formation of collapse pits. 243 244 Figure 8. Candidate ice-cored moraines. (a) Some moraines in the Ridged Facies near the Northwest Plateau (Scanlon et al., 2014a; see Figure 1) have troughs along their crests. CTX image mosaic. (b) Close view of the drop moraine with linear trough, highlighted by two arrows in Figure 8a. CTX image P17_007814_1773_XI_02S129W. a 245 b 246 Figure 9. Typical dunes with linear troughs, found throughout the western extent of the Arsia Mons fan-shaped deposit. CTX image P19_008605_1772_XI_02S129W. 247 Figure 10. The majority of the ridges with linear troughs, here interpreted as dunes, are oriented southwest-to-northeast wherever they occur, and are concentrated in local topographic lows rather than being associated with a specific stratigraphic level in the deposit. (a) CTX image mosaic. (b) CTX mosaic with dunes highlighted in red, drop moraines highlighted in blue, and white arrows indicating locations where dunes are concentrated in the local topographic lows between lava flows. a b 248 Figure 11. The dunes with linear troughs may have formed from dust-ice dunes (a). Upon a change from the climate in which they formed, ice would sublime from the surface of the dunes (b). Over time, this dry material would slump down the dune sides, exposing fresh ice-cemented material at the dune center (c) to continuing sublimation (d). a b c 249 d 250 Figure 12. The dunes with linear troughs may have formed from snow dunes (a) that gained a dust cover (b), e.g. from a pyroclastic eruption. Such a cover would slump away from the dune crest over time (c), forming cracks at the crest and exposing the snow beneath to sublimation (d). As the snow sublimed, dust would fall in to fill the space left behind (e). a b c d e 251 Figure 13. Dust- and sand-covered ripples in the McMurdo Dry Valleys. Boxes on left images indicate the areas shown in right images. (a) Small dust- and sand-covered snow ripples in the Dry Valleys develop cracks in the dust cover along their crests. (b) An alternate view of the debris-covered ripples shows bare ground between them. Dark central bars on scale card are 10 cm long. Photos taken by D. M. Hollibaugh Baker, 11/11/2010. a b 252 Figure 14. Topography of a region of pits (examples highlighted with white arrows) and knobs (examples highlighted with red arrows). Topographic contours from HRSC DEM, superimposed on DEM-shaded CTX imagery. Contour interval is 5 meters; bold contour every 20 m. 253 Chapter Five Synthesis Kathleen E. Scanlon and James W. Head Department of Earth, Environmental and Planetary Sciences, Brown University 324 Brook Street, Box 1846, Providence, RI, 02912, USA 254 The oldest terrains on Mars are marked by crater lake basins, often with breached rims that constrain the volume of water they held (e.g. Fassett and Head, 2008a), and by long and branching fluvial valleys (e.g. Hynek et al., 2010). Yet observed mineralogy suggests that little alteration occurred in most of these fluvial environments (e.g. Ehlmann et al., 2011; Goudge et al., 2012), and most warming mechanisms proposed so far are either insufficiently strong to induce widespread equatorial rainfall or snowmelt (e.g. Forget et al., 2013; von Paris et al., 2013; Kerber et al., 2015) or have not yet conclusively been demonstrated to be geologically plausible (e.g. Segura et al., 2008; Ramirez et al., 2014). The “icy highlands” model for Mars at the Noachian-Hesperian boundary (Wordsworth et al., 2013; 2015; Head and Marchant, 2014), in which ice accumulates in the highlands due to the altitude-dependence of surface temperature in a thicker atmosphere, but melts only periodically due to the limited greenhouse warming of a pure- CO2 atmosphere, is especially compelling for two reasons. First, it is a natural consequence of a thicker CO2 atmosphere, such as has been hypothesized for early Mars; unless the solar flux was not actually lower when the valley networks formed, or unless a geochemically plausible combination of greenhouse gases is determined that could warm the surface of Mars by several tens of degrees at just a few tens of millibars of surface pressure, the highlands of ancient Mars will have been substantially colder than the northern, Hellas, and Argyre basins. Second, the distribution of Noachian valley networks on Mars is better correlated to the distribution of ice predicted by 3D climate models on a cold early Mars than the distribution of rain on a warm, wet early Mars (Wordsworth et al., 2015). 255 Other endmember possibilities for the early Mars climate include: (1) A steady- state warm, wet early Mars. Reconciling such an environment with geomorphic evidence for intermittent fluvial activity (e.g. Barnhart et al., 2009; Buhler et al., 2014) and with evidence that few alteration minerals formed in surface fluvial environments may require that any continuously warm early Mars was very arid. Furthermore, true polar wander may be required to explain the distribution of valley networks if they formed primarily from rainfall rather than snowmelt (Wordsworth et al., 2015; Bouley et al., 2015), and as we showed in Chapter 2, the proposed form of true polar wander (Bouley et al., 2015) may not be consistent with the shape of the south polar Dorsa Argentea Formation (DAF) or the ages of landforms within it. (2) A cold and icy early Mars similar to the icy highlands model but with much lower surface pressure, such that heat transport is weak and the location of melting is more strongly controlled by the location of maximum insolation, which in turn is controlled by orbital parameters. Our results in Chapter 1 suggest that ~40 – 50°K additional greenhouse warming is required to achieve runoff rates consistent with observed channel widths in a 600 mb CO2 atmosphere, and our simulations at 25° obliquity with a 50 mb atmosphere suggest that maximum temperatures do not exceed 240°K in ice-covered equatorial highlands with a faint young sun even in an atmosphere at thin as 50 mb. In Scanlon et al. (2013), we showed that local-scale drainage density variations in several valley networks are consistent with orographic enhancement of precipitation in the prevailing wind directions expected for a cold early Mars with a dense (but < 2 bar) CO2 atmosphere. Wordsworth et al. (2013, 2015) showed that the stable large-scale ice distribution on a cold early Mars is correlated to valley network drainage density, and 256 Scanlon et al. (2013) showed that wind patterns would have enhanced ice accumulation in a manner consistent with observations at the local scale. Clearly ice could have migrated to the right locations to form the valley networks on a cold early Mars, but it must also be demonstrated that this ice could then melt at rates consistent with observations. In Chapter 1 of this work, we showed that runoff rates consistent with valley network interior channel morphometry (Irwin et al., 2005; Hoke et al., 2011) could consistently be reached on an early Mars with a 1000 mb CO2 atmosphere and additional greenhouse warming of > 30° and < 50° for most valleys. Furthermore, modeled snowmelt runoff rates were uniformly too high to be consistent with observed morphometry in climates with 40 – 50°K additional greenhouse warming, yet modeled rainfall rates in these climates were almost always well below the runoff rates inferred from observations. This indicates that snowmelt can generate sufficiently rapid runoff, and that snowmelt rates consistent with observations do not require as much warming as rainfall rates consistent with observations. Any model for the early Mars climate that explains the equatorial valley networks must also account for coeval features elsewhere on the planet in order to be valid. The central point of the icy highlands model is that temperatures on a cold Mars with a dense atmosphere are controlled by surface elevation. In Chapter 2, we called attention to the fact that when Tharsis, Hellas, and Argyre are present, this effect results in a south polar temperature minimum with the same asymmetric shape as the DAF, which has been interpreted as the remnants of an extensive south polar ice sheet dating to the same time as the equatorial valley networks (Head and Pratt, 2001; Kress and Head, 2015). An ice sheet with the full extent of the DAF only forms, and only undergoes the intense basal 257 melting inferred from large eskers and fluvial channels in and around the DAF, when temperatures are elevated by a greenhouse gas in addition to a thick CO2 atmosphere. Our results therefore support the DAF eskers having formed during the same warm intervals in the icy highlands model as the valley networks. An interesting component of our findings, however, is that when only CO2 is present (i.e. during cold intervals), ice accumulation in the highlands results in large ice sheets flowing into Hellas and Argyre basins from the east and southeast, with abundant basal melting occurring in both basins. Considering our results in Chapter 1, Chapter 2, and Scanlon et al. (2013) in the framework of the icy highlands model (Wordsworth et al., 2013; 2015; Head and Marchant, 2014), we can make three key geological predictions that can be used to test and refine the model. 1. Controls on age and development of valley networks. First, variations in cessation age and maturity between valley networks should be related to (a) predicted robustness of ice accumulation at their locations and (b) variations in peak surface temperatures between their locations in any given climate warming scenario. The icy highlands model proposes that most runoff on early Mars occurred due to snowmelt, and that this runoff occurred during a series of warming events, with water migrating back to the highlands as snow in between warm intervals. If later warming events were weaker than earlier events (such as might be expected if the warming was due to volcanism or impacts, both of which would have declined over time, or if warming events were superimposed upon a secular trend of CO2 loss), only the most easily warmed valley networks should have been active during later events. Valley network crater exposure ages appear to bear out this prediction to some extent (Fassett and Head, 2008b). Surface 258 temperatures are primarily a function of elevation in the icy highlands scenario, and higher-altitude valley networks such as Naro Vallis (which Fassett and Head dated to 3.81 Ga) and Licus Vallis (3.73 Ga) are in many cases substantially older than lower- altitude valley networks such as the populations including Paraná Valles (3.49 Ga) and Brazos Vallis (3.46 Ga). However, a few outliers exist (such as the unnamed high-altitude valley network at 132.5°E, 22°S, whose cessation is dated at 3.45 Ga), and as Fassett and Head (2008b) noted, the uncertainty in crater statistics when dealing with small areas such as those of the valley networks necessitates caution. Regardless of sequence, if some warming events were weaker than others, valley networks in more easily warmed locations should be better-developed and/or more deeply incised than those which require stronger warming to initiate snowmelt. Similarly, valley networks (such as those near Aonia Terra) where ice is only predicted to accumulate in a relatively narrow range of spin-axis obliquity conditions should be older and less well-developed than those where ice accumulation is predicted at all obliquities. In the “thin atmosphere, cold climate” paradigm for early Mars, lower-latitude valley networks might be expected to activate most easily and remain active longest, rather than lower-altitude valley networks. In a “steady-state warm” early Mars, the same might be true of valley networks closest to hypothetical surface sources of liquid water. If early Mars experienced true polar wander as the valley networks were forming, valley networks in the “Tharsis rain shadow” of Wordsworth et al. (2015) might be expected to have ceased before those where rainfall is also present with modern topography, unless valley network formation ceased entirely before true polar wander completed. Finally, as with valley network age and development, geomorphologic parameters such as lake 259 drainage ratio (Fassett and Head, 2008a) and best-fit “X-ratio” (e.g. Barnhart et al., 2009; Matsubara et al., 2013), which relate geomorphometry to climate aridity, should similarly be expected to vary differently across the surface of the planet depending on which of the aforementioned endmembers the early martian climate most resembled. 2. Hellas and Argyre ice sheets. Second, the Hellas and Argyre impact basins should contain evidence for wet-based glacial erosion of part or all of their walls, and glacial deposits on the basin floors. Our results from the UMISM glacial flow model in Chapter 2 show that at 42° obliquity, regardless of surface pressure, the accumulation of ice on the highlands surrounding Hellas and Argyre basins results in wet-based glacial flow into the basins from the east at Hellas and the southeast at Argyre. This prediction is supported by observed geology. The southeastern portion of the floor of Hellas basin is covered with loosely consolidated, voluminous deposits dating to the Noachian-Hesperian boundary and interpreted as having been eroded by glacial and fluvial activity from the basin rim and volcanic plains to the south and east of the basin and altered in aqueous environments (Bernhardt et al., 2016). Bernhardt et al. (2016) interpreted the trough along the northwestern edge of the basin floor as an aeolian erosional feature, but we propose in light of UMISM results that the glaciofluvial deposits may never have extended completely to the northwestern basin wall. Similarly, lobate material extending northwest from the southeastern edge of Argyre has been interpreted as material eroded from the Argyre rim and ejecta by glacial and fluvial activity in the late Hesperian (Dohm et al., 2015) or as glacial deposits from the early Hesperian (Bernhardt et al., 2013). No evidence is present for Noachian-Hesperian glacial activity along the west rim of Hellas, as is seen in UMISM simulations with high 260 global ice inventories. Basal melting is weak and scattered in the modeled west-rim ice sheet, however; if such an ice sheet would have been mostly cold-based, a lack of evidence for it may not necessarily indicate that it never existed. Our second prediction invites several questions for future investigation. First, is there any evidence for banding in the composition or thermal properties of the hypothesized glacial deposits? If deposition of these ice sheets is a robust feature of the “cold state” of the icy highlands model, and melting of any ice in these basins is expected to occur readily in the “warm state”, the glacial deposits might be expected to record multiple glacial advances. The final cycles of wet-based glaciation and lacustrine activity could, however, have obliterated any record of earlier cycles. Second, is there any evidence for glaciotectonic deformation “downstream” of the hypothesized Hellas and Argyre glacial flows? Proglacial lacustrine sediments should be readily deformable by glacial thrusting, and the hypothesis (Moore and Wilhelms 2001; 2007) that the concentric ridges of the Hellas “honeycomb terrain” formed by glacial deformation of lacustrine sediments should be revisited. Third, what is the modeled transport time of ice between the rims of Hellas and Argyre basins (where abundant ice accumulation occurs in the “cold state” of the icy highlands model) and the Tharsis plateau (where ice is most stable in the “warm state”)? Combined with constraints on warm intervals from minimum filling times of open-basin lakes, this time scale could provide a constraint on the minimum length of cold intervals. 3. Hellas and Argyre oceans. The third major prediction that follows from our results is that lacustrine deposits should be present in both Hellas and Argyre basins. Due to the depth of the basins, these regions experience annual average surface temperatures 261 above freezing when the equatorial highlands are 20-30° colder (Chapter 1), and ice sheets in the basins experience widespread basal melting even in climates with no greenhouse gases other than CO2 (Chapter 2). Unless the early Mars atmosphere was very thin, in any climate warm enough for abundant rainfall or snowmelt in the equatorial highlands, Hellas and Argyre should have been warm even by terrestrial standards. Several workers have interpreted Noachian and Noachian-Hesperian units within these basins as consistent with paleolakes in the basins (e.g. Hiesinger and Head, 2002; Dohm et al., 2015; Bernhardt et al., 2016). In the “cold Mars, thin atmosphere” endmember model for the early Mars climate, where atmospheric pressure has always been low, temperatures in the Hellas and Argyre basins would not be expected to exceed those in the highlands as much as in the icy highlands model; evidence that there were no long- standing bodies of water in these basins would therefore support that model. If early Mars did resemble the icy highlands model, our results suggest that standing bodies of water would have been present in the Hellas and Argyre basins throughout the periods of fluvial activity in the equatorial valley networks and open-basin lakes, and that expansive wet-based ice sheets would have flowed into the basins during cold periods between episodes of fluvial activity in the highlands. In the icy highlands paradigm, therefore, the longest-lived aqueous environments on early Mars may have been in these basins, and their Noachian-Hesperian floor units should be high-priority targets for future astrobiological exploration. In addition to testing the icy highlands model, there are several other lines of investigation that would be particularly useful for increasing the detail in our picture of early Mars. 262 First, more terrestrial and modeling work is needed to seek out morphometric differences between valley networks that formed in very cold climates and those that formed primarily from rain in more temperate climates. Bogaart et al. (2003) showed that channel networks formed in periglacial conditions are “expanded” (i.e. have more, and higher-altitude, low-order tributaries) relative to those in the same catchment during temperate conditions. This observation, and their predictions for sediment yield variations in catchments cycling between temperate and permafrost conditions have interesting implications for Mars, and might be extended to produce hypotheses about channel morphometry and lacustrine sedimentation for testing the icy highlands model. Head and Cassanelli (2014) predicted that due to the relatively high erodibility of the ice table relative to dry soil above it, deepening of valley networks during initial incision into permafrost should be slower than broadening, and that the lowering of the ice table during cold intervals should favor re-use of the same channels, resulting in deep channels with low drainage density. Finally, landscape evolution model studies for Mars thus far have generally assumed constant precipitation throughout the catchment (e.g. Howard, 2007; Barnhart et al., 2009; but see also Matsubara et al., 2013). In the context of the icy highlands model, runoff would be expected to originate at the bottom of permeable layers on glaciers, and would be unable to erode the underlying regolith (and hence unable to affect the geologic record) until reaching the edge of the ice sheet. The branching parameters and changes in dimension downstream for valleys formed this way should therefore differ significantly from valleys formed in more temperate conditions, where constant runoff across the catchment is a more reasonable assumption. 263 Second, even if it can be shown that the morphology, morphometry, and distribution of fluvial and glacial features on Mars is most consistent with the icy highlands model, the mechanism for inducing warm intervals remains unknown. Much of the work that must be done to solve this problem falls to geochemistry (e.g. Ramirez et al. 2014; Batalha et al., 2015), but one test might be accomplished from sedimentology alone. Both impacts and volcanic eruptions have been proposed as triggers for warm intervals on early Mars, and in addition to greenhouse gases, both mechanisms would be expected to provide abundant loose sediment in the form of volcanic ash or impact ejecta. Evidence that layers in lacustrine sediments near volcanoes or impact craters hypothesized to have been agents of warming at the Noachian-Hesperian boundary begin with material compositionally consistent with volcanic ash or impact ejecta, before grading into local catchment material, would be compelling evidence for either mechanism. Third, more knowledge of the formation mechanisms for comparably large terrestrial eskers would improve our ability to draw conclusions from the eskers in Argyre basin and the Dorsa Argentea Formation. The formation of several small terrestrial eskers has been observed (e.g. Russell et al., 2001; Burke et al., 2008), and the sedimentology of several small esker systems has been studied in detail. With the exception of single-event eskers (e.g. Burke et al., 2012), the majority of studies point to a common paradigm of time-transgressive esker formation near the edges of retreating ice sheets, with surface melt as a dominant input of water to the system (e.g. Banerjee and MacDonald, 1975; Mäkinen, 2003; Hooke & Fastook, 2007; Storrar et al., 2014). Our results suggest that basal melt may have been the primary source of the water that ran 264 through subglacial channels in the DAF ice sheet; determining whether similarly large terrestrial eskers also required a substantial basal component, and whether sedimentological evidence for time-transgressive formation is as prevalent for eskers that formed in large part from basal melting as it is for primarily surface melt-fed eskers, would be of value for future investigations. In Scanlon et al. (2014), we documented numerous landforms indicative of volcano-ice interaction in the Arsia Mons fan-shaped deposit, and estimated the size and longevity of liquid water bodies that would result from these glaciovolcanic interactions. In Chapter 3, we demonstrated evidence that this glaciovolcanic activity resulted in local wet-based glacial sliding in parts of the Arsia Mons FSD, counter to the dominant cold- based glacial paradigm of the Amazonian era on Mars, and in Chapter 4, we presented geomorphological evidence that some of this ~200 Ma ice may still be present in the deposit. One of the most interesting aspects of glaciovolcanic landforms is their capacity to provide long-lived and voluminous (by Amazonian standards) surface bodies of liquid water, and hence possible oases for any martian microbial life. Even terrestrial glaciovolcanic environments are still relatively little-studied as habitats, however, and several avenues of future research would be particularly helpful in maximizing their usefulness as analogues for Mars. First, more descriptions of chemolithoautotrophy in glaciovolcanic microbial ecosystems would help reinforce the case for these environments’ practical habitability on Mars. In many terrestrial subglacial environments, oxygen from glacial ice and legacy organic carbon are important ecosystem drivers, and neither is likely to have been abundant in martian analogues. Second, more descriptions 265 of how these environments are colonized on Earth would be helpful in determining the likelihood of glaciovolcanic environments on Mars becoming inhabited if dormant martian microbial life existed to colonize them. In our warmest GCM simulations, those where rainfall occurs in some areas and snowmelt rates exceed runoff rates inferred for the valley networks, the most stable location for ice is in the highest-altitude regions of Mars, primarily the Tharsis volcanic plateau. Given the hypothesis that valley-forming warm intervals were induced by greenhouse gases released during episodes of volcanic activity, particularly in Tharsis, this raises the intriguing possibility of widespread volcano-ice interactions in the region at the Noachian-Hesperian boundary. Any volcano-ice interactions at Tharsis could also have resulted in substantial infiltration of surface water to subsurface aquifers. Evidence for or against this interaction should be sought out in the geologic record. One potential starting point is the prominent scarp surrounding Olympus Mons; Hodges and Moore (1979) proposed that the scarp might be analogous to the steep sides of Icelandic table mountains, suggesting that kilometers of ice surrounded the volcano in the early stages of its growth; De Blasio (2012) also argued that contact cooling by water was necessary to allow the steep scarp to form. Helgason (1999) also argued that the scarp was glaciovolcanic in origin, though in his interpretation the scarp is not a direct lava-ice contact feature but rather a collapse feature caused by the instability of subaerial lavas emplaced atop ice. Furthermore, the aureole deposits surrounding Olympus Mons have been proposed as flow features assisted by ground ice (e.g. Tanaka, 1985). In the work we have completed during my education at Brown, as in many of the key papers this thesis builds on (Forget et al., 2006; Fastook et al., 2008; Fastook et al., 266 2012; Wordsworth et al., 2015), we combined surveys of geomorphology with information from both new and pre-existing GCM simulations to constrain the warmth and surface pressure of Mars as the valley networks were forming (Scanlon et al., 2013, Chapter 1, and Chapter 2), add to the inventory of aqueous environments on recent Mars (Scanlon et al., 2014, and Chapter 3), and add to the inventory of present-day equatorial ice on Mars (Chapter 4). Detailed sedimentological and geomorphometric surveys will be required to further validate, refine or refute the icy highlands model, or any other paradigm for the climate of early Mars. Using these geologic results to distinguish possibilities from among the multitude of scenarios in GCM parameter space is an approach that provides more returns than sum of its parts. There is much work to be done, and I look forward to starting. References Banerjee, I., & McDonald, B. C. l975, Nature of esker sedimentation. Glaciofluvial and Glaciolacustrine Sedimentation: Society of Economic Paleontologists and Mineralogists Sp. Pub, (23), l32-l54. Barnhart, C. J., Howard, A. D., & Moore, J. M. (2009). Long‐term precipitation and late‐ stage valley network formation: Landform simulations of Parana Basin, Mars. Journal of Geophysical Research: Planets, 114(E1). Batalha, N., Domagal-Goldman, S. D., Ramirez, R., & Kasting, J. F. (2015). Testing the early Mars H2–CO2 greenhouse hypothesis with a 1-D photochemical model. Icarus, 258, 337-349. Bernhardt, H., Hiesinger, H., Ivanov, M. A., Ruesch, O., Erkeling, G., & Reiss, D. (2016). 267 Photogeologic mapping and the geologic history of the Hellas basin floor, Mars. Icarus, 264, 407-442. Bogaart, P. W., Tucker, G. E., & De Vries, J. J. (2003). Channel network morphology and sediment dynamics under alternating periglacial and temperate regimes: a numerical simulation study. Geomorphology, 54(3), 257-277. Bouley, S., Baratoux, D., Matsuyama, I., Forget, F., Costard, F., & Séjourné, A. (2015). True Polar Wander Recorded by the Distribution of Martian Valley Networks. Lunar and Planetary Science Conference 46, abstract #1887. Buhler, P. B., Fassett, C. I., Head, J. W., & Lamb, M. P. (2014). Timescales of fluvial activity and intermittency in Milna Crater, Mars. Icarus, 241, 130-147. Burke, M. J., Woodward, J., Russell, A. J., Fleisher, P. J., & Bailey, P. K. (2008). Controls on the sedimentary architecture of a single event englacial esker: Skeiðarárjökull, Iceland. Quaternary Science Reviews, 27(19), 1829-1847. Burke, M. J., Brennand, T. A., & Perkins, A. J. (2012). Transient subglacial hydrology of a thin ice sheet: insights from the Chasm esker, British Columbia, Canada. Quaternary Science Reviews, 58, 30-55. De Blasio, F. V. (2012). Conceptual model for the origin of the Olympus Mons cliffs, Mars: An essential influence of water?. Planetary and Space Science, 69(1), 105- 110. Dohm, J. M., Hare, T. M., Robbins, S. J., Williams, J. P., Soare, R. J., El-Maarry, M. R., Conway, S. J., Buczkowski, D. L., Kargel, J. S., Banks, M. E., Fairén, A. G., Schulze-Makuch, D., Komatsu, G., Miyamoto, H., Anderson, R. C., Davila, A. F., Mahaney, W. C., Fink, W., Cleaves, H. J., Yan, J., Hynek, B., & Maruyama, S. 268 (2015). Geological and hydrological histories of the Argyre province, Mars. Icarus, 253, 66-98. Ehlmann, B. L., Mustard, J. F., Murchie, S. L., Bibring, J. P., Meunier, A., Fraeman, A. A., & Langevin, Y. (2011). Subsurface water and clay mineral formation during the early history of Mars. Nature, 479(7371), 53-60. Fassett, C. I., & Head, J. W. (2008a). Valley network-fed, open-basin lakes on Mars: Distribution and implications for Noachian surface and subsurface hydrology. Icarus, 198(1), 37-56. Fassett, C. I., & Head, J. W. (2008b). The timing of Martian valley network activity: Constraints from buffered crater counting. Icarus, 195(1), 61-89. Fastook, J. L., Head, J. W., Marchant, D. R., & Forget, F. (2008). Tropical mountain glaciers on Mars: Altitude-dependence of ice accumulation, accumulation conditions, formation times, glacier dynamics, and implications for planetary spin-axis/orbital history. Icarus, 198(2), 305-317. Fastook, J. L., Head, J. W., Marchant, D. R., Forget, F., & Madeleine, J. B. (2012). Early Mars climate near the Noachian–Hesperian boundary: Independent evidence for cold conditions from basal melting of the south polar ice sheet (Dorsa Argentea Formation) and implications for valley network formation. Icarus, 219(1), 25-40. Forget, F., Haberle, R. M., Montmessin, F., Levrard, B., & Head, J. W. (2006). Formation of glaciers on Mars by atmospheric precipitation at high obliquity. Science, 311(5759), 368-371. Forget, F., Wordsworth, R., Millour, E., Madeleine, J. B., Kerber, L., Leconte, J., Marcq, E., & Haberle, R. M. (2013). 3D modelling of the early martian climate under a 269 denser CO2 atmosphere: Temperatures and CO2 ice clouds. Icarus, 222(1), 81-99. Goudge, T. A., Head, J. W., Mustard, J. F., & Fassett, C. I. (2012). An analysis of open- basin lake deposits on Mars: Evidence for the nature of associated lacustrine deposits and post-lacustrine modification processes. Icarus, 219(1), 211-229. Head, J. W., & Pratt, S. F. (2001), Extensive Hesperian-aged south polar ice sheet on Mars: Evidence for massive melting and retreat, and lateral flow and ponding of meltwater, Journal of Geophysical Research: Planets, 106 (E6), 12,275-12,299. Head, J. W., & Cassanelli, J. P. (2014). Valley network formation: predictions for fluvial processes in a late Noachian icy highland climate regime. In Lunar and Planetary Science Conference 45, abstract #1413. Head, J. W., and Marchant, D. R. (2014), The climate history of early Mars: Insights from the Antarctic McMurdo Dry Valleys hydrologic system, Antarctic Science, 26, 774-800. Helgason, J. (1999). Formation of Olympus Mons and the aureole-escarpment problem on Mars. Geology, 27(3), 231-234. Hiesinger, H., & Head, J. W. (2002). Topography and morphology of the Argyre Basin, Mars: implications for its geologic and hydrologic history. Planetary and Space Science, 50(10), 939-981. Hodges, C. A., & Moore, H. J. (1979). The subglacial birth of Olympus Mons and its aureoles. Journal of Geophysical Research: Solid Earth, 84(B14), 8061-8074. Hooke, R. L., & Fastook, J. (2007). Thermal conditions at the bed of the Laurentide ice sheet in Maine during deglaciation: implications for esker formation. Journal of Glaciology, 53(183), 646-658. 270 Howard, A. D. (2007). Simulating the development of Martian highland landscapes through the interaction of impact cratering, fluvial erosion, and variable hydrologic forcing. Geomorphology, 91(3), 332-363. Hynek, B. M., Beach, M., & Hoke, M. R. (2010). Updated global map of Martian valley networks and implications for climate and hydrologic processes. Journal of Geophysical Research: Planets, 115(E9). Kerber, L., Forget, F., & Wordsworth, R. (2015). Sulfur in the early martian atmosphere revisited: Experiments with a 3-D Global Climate Model. Icarus, 261, 133-148. Kress, A. M., & Head, J. W. (2015), Late Noachian and early Hesperian ridge systems in the south circumpolar Dorsa Argentea Formation, Mars: Evidence for two stages of melting of an extensive late Noachian ice sheet, Planetary and Space Science, 109-110, 1-20. Mäkinen, J. (2003). Time‐transgressive deposits of repeated depositional sequences within interlobate glaciofluvial (esker) sediments in Köyliö, SW Finland. Sedimentology, 50(2), 327-360. Matsubara, Y., Howard, A. D., & Gochenour, J. P. (2013). Hydrology of early Mars: Valley network incision. Journal of Geophysical Research: Planets, 118(6), 1365-1387. Moore, J. M., & Wilhelms, D. E. (2001). Hellas as a possible site of ancient ice-covered lakes on Mars. Icarus, 154(2), 258-276. Moore, J. M., & Wilhelms, D. E. (2007). Geologic Map of Part of the Western Hellas Planitia, Mars (No. 2953). Geological Survey (US). Storrar, R. D., Stokes, C. R., & Evans, D. J. (2014). Morphometry and pattern of a large 271 sample (> 20,000) of Canadian eskers and implications for subglacial drainage beneath ice sheets. Quaternary Science Reviews, 105, 1-25. Ramirez, R. M., Kopparapu, R., Zugger, M. E., Robinson, T. D., Freedman, R., & Kasting, J. F. (2014). Warming early Mars with CO2 and H2. Nature Geoscience, 7(1), 59-63. Russell, A. J., Knudsen, O., Fay, H., Marren, P. M., Heinz, J., & Tronicke, J. (2001). Morphology and sedimentology of a giant supraglacial, ice-walled, jökulhlaup channel, Skeiðarárjökull, Iceland: implications for esker genesis. Global and Planetary Change, 28(1), 193-216. Scanlon, K. E., Head, J. W., Madeleine, J. B., Wordsworth, R. D., & Forget, F. (2013). Orographic precipitation in valley network headwaters: Constraints on the ancient Martian atmosphere. Geophysical Research Letters, 40(16), 4182-4187. Scanlon, K. E., Head, J. W., Wilson, L., & Marchant, D. R. (2014). Volcano–ice interactions in the Arsia Mons tropical mountain glacier deposits. Icarus, 237, 315-339. Segura, T. L., Toon, O. B., & Colaprete, A. (2008). Modeling the environmental effects of moderate‐sized impacts on Mars. Journal of Geophysical Research: Planets, 113(E11). von Paris, P., Grenfell, J. L., Rauer, H., & Stock, J. W. (2013). N 2-associated surface warming on early Mars. Planetary and Space Science, 82, 149-154. Wordsworth, R., Forget, F., Millour, E., Head, J. W., Madeleine, J. B., & Charnay, B. (2013). Global modelling of the early martian climate under a denser CO2 atmosphere: Water cycle and ice evolution. Icarus, 222(1), 1-19. 272 Wordsworth, R. D., Kerber, L., Pierrehumbert, R. T., Forget, F., & Head, J. W. (2015). Comparison of “warm and wet” and “cold and icy” scenarios for early Mars in a 3‐D climate model. Journal of Geophysical Research: Planets, 120(6), 1201-1219. 273