The performance of CMIP6 models in simulating surface energy fluxes over global continents

Correct simulation of land surface energy balance is critically important in reliably projecting future climate change. In this study, simulated surface energy fluxes including net radiation, latent heat, and sensible heat from 22 CMIP6 models are evaluated against ERA5 reanalysis data during 1950–2014 over global land surface to assess the performance of CMIP6 models in simulating surface energy balance. Results reveal significant differences in seasonal and regional characteristics of CMIP6 model simulations in the surface energy flux predictions. Net radiation is slightly overestimated in JJA (June, July, August) but underestimated in DJF (December, January, and February). Most models overestimate net radiation in both seasons over Africa, Australia, and Latin America. The inter-model variability of biases is substantial in DJF net radiation simulations over Asia, Europe and Siberia, and North America, although there is a good agreement between simulations of selected models and the ERA5 reanalysis data in JJA except for Greenland. Moreover, models overestimate latent heat and underestimate sensible heat in DJF, although latent heat and sensible heat simulations match well with the reference values in JJA. Latent heat is overestimated in DJF over Australia, Europe and Siberia, and Latin America. Sensible heat is underestimated during DJF over Europe and Siberia and Latin America. The performance of CMIP6 in Rn simulation is better than LH simulation and SH simulation for both JJA and DJF. Findings from this study will provide useful references for surface energy data users and future model development.