Databases for sea surface paleotemperature based on geochemical proxies from marine sediments: Implications for model-data comparisons

Abstract

The Paleoclimate Model Intercomparison Project (PMIP) aims to simulate the response of Earth's climate system to changes in different climate forcing factors, focusing on six key intervals of the geological past (the mid-Piacenzian Warm Period at ∼3.3-3.0 Ma, the last interglacial period at ∼ 127 ka, the last glacial maximum at ∼21 ka, the mid-Holocene at ∼6 ka, the last millennium i.e. the 850-1850 Common Era (CE) time window, as well as the last glacial termination, i.e. the ∼21-9 ka time window). Model evaluation is ultimately achieved via model-data comparison exercises, and such diagnoses rely on databases of proxy-based paleoclimate reconstructions available in the literature. In this context, we provide a review of the main geochemical sea surface temperature (SST) proxies (namely foraminiferal Mg/Ca ratio, the UK'37, and the TEX86) that are routinely used to aliment databases used in model-data comparisons, focusing on proxy systematics and specificities that may affect or bias SSTs reconstructed with each of these proxies. We then go on to discuss notorious mismatches between SST estimates derived from different proxies for all periods studied in PMIP, and aim to provide guidance for a better interpretation of these model-data mismatches. This review is deliberately aimed at helping the paleoclimate modeling community to better appraise how information on past SST variability is transferred to the marine geological record. In particular, when SST estimates based on different proxies provide contrasting SST signals, we stress the importance of looking at those contrasting signals through the prism of the ecology of the organisms at the origin of the SST proxies.