Application of Global Paleovegetation Data for Benchmarking Paleoclimate Simulations

Climate models provide an opportunity for testing hypotheses concerning the causes of past climatic variations, providing the consistent explanations of past climate changes, and simulating potential future climate changes. We compare paleovegetation syntheses from the Last Glacial Maximum (LGM) and Mid Holocene (MH) with simulations performed as part of the Paleoclimate Modelling Intercomparison Project (PMIP) 2 and 3 projects, with the aim of evaluating the ability of the simulations to reproduce the key regional and global patterns of climate recorded by the data. We apply two approaches using vegetation models in a data-model comparison framework: a forward-modeling approach that simulates vegetation using climate-model output, and an inverse-modeling approach that uses the vegetation data to infer the past values of the specific climate that controlled vegetation distributions. In the forward-modeling approach, we use the BIOME 4 equilibrium-biogeochemistry model and palaeovegetation data (e.g. BIOME 6000), which includes new regional pollen data sets from Australia, Southeast Asia, South America, and the Indian subcontinent, in order to evaluate the response of PMIP2 AOGCMs simulations of the LGM and MH. For the inverse approach, we use BIOME4 to iteratively estimate the potential paleoclimate consistent with based on the BIOME6000 data using a Monte Carlo Markov Chain algorithm. BIOME 4 employs mechanistic descriptions of the relationship of vegetation on climate and also allows the direct effects of carbon dioxide concentration to be considered. In addition to standard (i.e. map-comparison) approaches for comparing the simulated/observed vegetation and climate, we show some diagnostics based the mapping of observed and simulated biomes in climatic spaces. These diagnoses can provide information about the specific climatological explanations for the mismatches between the simulations and observations.