A41E-0109
Data Assimilation-Based SST Reconstruction from Coral δ18Ο Records

Thursday, 17 December 2015
Poster Hall (Moscone South)
Maud Comboul1, Julien Emile-Geay1, Gregory J Hakim2 and Michael N Evans3, (1)University of Southern California, Los Angeles, CA, United States, (2)University of Washington, Seattle, WA, United States, (3)University of Maryland College Park, College Park, MD, United States
Abstract:
Reconstructing past spatio-temporal climate variability requires quality paleo-proxy data with broad spatial coverage. Yet those archives are noisy recordings of their environment and are always uncertain to some degree. Given a network of paleoclimate observations, we propose a data assimilation (DA) framework that allows us to reconstruct past climate along with its uncertainty estimate. Starting with a distribution of climate states of interest given by an ensemble of instrumental or simulated observations, and a realistic Proxy System Model (PSM) to connect the climate representation to the paleoclimate records, our method assimilates the proxy data to update the target distribution. The believed degree of uncertainty present in the archives is explicitly modeled via the PSM and is therefore inherently propagated throughout the reconstruction process into the informed climate distribution.
We illustrate our method with the reconstruction of past sea surface temperature (SST) fields from a network of coral δ18Ο records. Noisy SST fields extracted from the Community Climate System Model 4.0 Last Millennium simulation and a pseudproxy network of coral δ18Ο are used as a proof of concept to show how the ratio between the noise present in the fields and the paleoclimate observation uncertainty affect the quality of the reconstructions.