Surface Energy Partitioning Biases in Major Reanalyses Revealed by Soil Moisture Drying Dynamics

Abstract

Identifying the flux partition regimes between soil moisture (SM) and evaporative fraction (EF, ratio of evapotranspiration to available energy) is important for understanding the hydrometeorological processes as well as the development of Land Surface Models (LSMs). However, evaluating the SM‐EF regimes (i.e., whether EF is controlled by energy or water availability) in LSMs remains challenging since concurrent large‐scale observations of the diagnostic variables are often scarce. Recent advancements in satellite techniques provide unique advantages for evaluating the models' performance. Here, we use long‐term satellite SM data sets NNSM and observation‐based meteorological forcing data sets to evaluate the SM‐EF regime in six major reanalyses (i.e., GLDAS‐Noah, GLDAS‐CLSM, MERRA2, NCEP‐FNL, ERA5, and JRA5). The analyses are conducted over China and North America. The results show that at large scale, all data sets consistently overestimate the degree of water limitation, though the bias varies across models and largely depends on how each parameterizes soil and vegetation controls on surface water–energy partitioning. The models' land‐atmosphere coupling configurations only show moderate influence on EF regimes, and the effect of surface soil layer thickness is minimal. The comparison of model behaviors between the two study regions further reveals that inter‐model discrepancies are much more pronounced in China than in North America. This reflects differences in data quality and model calibration density between the two regions. Our study therefore advances our understanding in land surface models' hydrometeorological behaviors and provides valuable reference to further improve the model parameterization on surface energy partitioning processes.