Climate Models Underestimate Satellite Era Land-ocean Warming Contrast in the Tropics

General circulation models (GCM) can reasonably reproduce the global mean temperature trend during the historical period. In this study, we examine the performance of CMIP6 models in reproducing the changes in land-ocean temperature contrast between 1979-2014 during the comprehensive satellite observation era. The observed land-ocean warming contrast, defined as the land warming trend divided by the ocean warming trend, is completely outside the model spread of the historical scenario, indicating that the models severely underestimate land temperature increase relative to global mean temperature warming. Even when sea surface temperatures are prescribed to observations in AMIP experiments, the land warming trend remains outside the model spread, particularly between 15S and 15N. This was shown to be because GCM overestimates the increase in specific humidity on tropical land and underestimates the drying trend on tropical land. Since future projections over land have a significant impact on human activity, improving the representation of tropical land surface processes in GCMs is essential.