Decomposition and Attribution of Land Surface Temperature Bias Over China in CoLMv2024 and Noah‐MP Land Surface Model

Land surface temperature (Ts) is crucial for understanding surface energy balance and climate change. This study investigates the sources of Ts biases in CoLMv2024 and Noah‐MP land surface model (LSM) using identical forcing data over China by decomposing the surface energy balance equation using ERA5‐Land reanalysis (ERA5L), MODIS, and ground‐based reference data sets. Both LSMs show significant seasonal and regional Ts biases, characterized by summer warm biases and pronounced winter cold biases. Uncertainty in energy fluxes across reference data sets limits single‐dataset evaluations. Energy fluxes in both LSMs are generally within uncertainty ranges of reference data, except that Noah‐MP consistently underestimates winter net radiation and sensible heat flux. Compared with ERA5L, the largest Ts bias across regions is caused by radiation forcing, that is, downward solar radiation or longwave radiation (DLW), except that the largest winter cold bias is caused by upward solar radiation (USW) in Noah‐MP in Northern China. Compared with MODIS, DLW probably contributes most to Ts biases for Noah‐MP in autumn and for CoLMv2024 in autumn and winter, except that USW contributes most to the winter cold bias in Noah‐MP. In March and April, the largest contributing term is ground heat flux in both LSMs. Despite large discrepancies in simulated snow properties, Ts simulations in two LSMs over the snow‐covered regions are similar and underestimated during the cold season. This study emphasizes the importance of radiation forcing data and the parameterization of processes other than snow physics in complex regions in cold seasons.