Numerical climate models are sophisticated tools used to predict the climate of the future. To be confident in their predictions, model simulations of the modern climate are usually compared with observational data from the last 50 – 100 years, the time period for which instrumental records exist.
Integrating Ocean Biogeochemistry, Paleoceanography and Climate Modeling
Another more rigorous test, however, is to apply the models to the climate of the geologic past, for which the boundary conditions are sufficiently known. The aim is to test whether the models are flexible enough to reproduce climate conditions distinctly different from today.
A multidisciplinary team combining expertise in ocean biogeochemical modeling, paleoceanography, and climate modeling achieved an unprecedented model-data synthesis of sea surface temperature (SST) changes over the last 10 kyrs (Holocene). Using paleo proxy SST -data based on (i) marine phytoplankton and zooplankton organisms, (ii) modern satellite data of SST and phytoplankton productivity, and (iii) a Holocene simulation of the Kiel Climate Model (KCM), it was shown that paleo-proxies may preferentially record seasonal signals instead of annual mean climate conditions, which is the usual interpretation. Furthermore, the respective seasons may differ between regions (see Fig.). After taking seasonal preferences into account in model-proxy comparisons, the model-data agreement was considerably improved; however, the climate model still systematically underestimated the Holocene SST trends (see Fig.).
Schneider, B., Leduc, G., Park, W. (2010) Disentangling seasonal signals in Holocene climate trends by satellitemodel-proxy integration. Paleoceanography 25, PA4217, doi: 10.1029/2009PA001893.
Caption: a) Climate signal that yields the best match for modeled and reconstructed Holocene SST trends for the phytoplankton (UK‘37) proxy data; green: SST weighted by the monthly primary productivity as obtained from the satellite data; grey: annual mean SST; red: warmest month of the year; blue: coldest month of the year. The larger the symbol the better the match between model and proxy data. b) Seasonally sub-sampled SST trends from the model (colored dots) yield a better match with proxy data than annual mean SSTs from the model (open circles), although some systematic discrepancy remains.