Improving biome and climate modelling for a set of past climate conditions: evaluating bias correction using the CDF-t approach

Abstract Climate model simulations are inherently biased. It is a notably difficult problem when dealing with climate impact assessments and model-data integration. This especially true looking at derived quantities such as biomes, where not only but also vegetation dynamics biases come into play. To overcome difficulties, we evaluate the performance of an existing methodology to correct outputs, applied here for first time long past conditions. The proposed relies on “Cumulative Distribution Function - transform” (CDF-t) technique, which allows account change within bias-correction procedure. results evaluated in two independent ways: i- using forward modelling, so that directly comparable reconstructed distribution; ii- climatic reconstructions based inverse modelling approach. performed intermediate complexity iLOVECLIM standard global interactively downscaled over Europe version. combined effects dynamical downscaling bias correction resulted significantly stronger agreement between simulated pollen-based biome (BIOME6000) pre-industrial (0.18 versus 0.44) mid-Holocene (0.31 0.40). Higher correlation observed statistically modelled gridded potential natural distribution biomes (0.36 0.41). Similarly, find higher temperatures (0.02 0.21). No significant difference found Last Glacial Maximum temperature reconstructions, due low number data points available. Our findings show application CDF-t method variables enables us simulate palaeoclimate better reconstructions.