Bayesian Inference in Statistical Analysis of Paleoclimate Records Ahn, Seonmin creator 2016 xiv, 183 p. born digital Thesis (Ph.D. -- Brown University (2016) Lawrence, Charles Director Herbert, Timothy Reader Lisiecki, Lorraine Reader Brown University. Applied Mathematics sponsor theses Studies in paleoclimatology should be accompanied by uncertainty analyses, as uncertainty is found within the very first measurements of the proxies imprinted in geological records. From there, due to the nature of measurements and the complexity of the methods, uncertainty continues to increase throughout the process of analysis. Complete descriptions of paleoclimate records are achieved only through using algorithms that consider uncertainties that are inherent to both paleoclimate records and methods of analysis. In this thesis, we introduce three statistical tools based on Bayesian inference to analyze paleoclimate records and present results using the three tools. First, we construct a probabilistic stack from globally distributed benthic δ18O records using a profile hidden Markov model. Next, we investigate the relative timings of glacial terminations between the eastern and western tropical Pacific by combining three statistical tools: Bayesian calibration, HMM-Match algorithm, and Bayesian change point algorithm. Lastly, we analyze multidimensional paleoclimate proxies measured from the core at the central Peru margin to obtain their cross-correlation and underlying state changes using the Kalman filter and the hidden Markov autoregressive model. we expect the statistical tools that we introduce in this thesis to play a critical role in rigorously analyzing paleoclimate data. Bayesian statistical analysis Paleoclimate analysis RPB 20160629 engEnglish10.7301/Z0DN43GCIn CopyrightCollection is open for research.dissertations