Abstract

Stratigraphic alignment is the primary way in which long marine climate records are placed on a common age model. We previously presented a probabilistic pairwise alignment algorithm, HMM-Match, which uses hidden Markov models to estimate alignment uncertainty and apply it to the alignment of benthic δ18O records to the" LR04" global benthic stack of Lisiecki and Raymo (2005)(Lin et al., 2014). However, since the LR04 stack is deterministic, the algorithm does not account for uncertainty in the stack. Here we address this limitation by developing a probabilistic stack, HMM-Stack. In this model the stack is a probabilistic inhomogeneous hidden Markov model, aka profile HMM. The HMM-stack is represented by a probabilistic model that" emits" each of the input records (Durbin et al., 1998). The unknown parameters of this model are learned from a set of input records using the expectation maximization (EM …

Date

January 1, 1970

Authors

Lorraine E Lisiecki, Seonmin Ahn, Deborah Khider, Charles Lawrence

Journal

AGU Fall Meeting Abstracts

Volume

2015

Pages

PP13D-07