Hyper-resolution land surface modeling provides an unprecedented opportunity to simulate locally relevant water and energy cycles. However, the available meteorological forcing data is often insufficient to fulfill the requirement of hyper-resolution modeling. Here, we developed a comprehensive downscaling framework based on topography-adjusted methods and automated machine learning (AutoML). With this framework, a 90 m atmospheric forcing dataset is developed from ERA5 data at a 0.25° resolution, and the Common Land Model (CoLM) is then forced with the developed forcing data over two complex terrain regions (Heihe and Upper Colorado River basins). We systematically evaluated the downscaled forcing and the CoLM outputs against both in-situ observations and gridded data. The ground-based validation results suggested consistent improvements for all downscaled forcing variables. The downscaled forcings, which incorporated detailed topographic features, offered improved magnitude estimates, achieving a comparable level of performance to that of regional reanalysis forcing data. The downscaled forcing driving the CoLM model show comparable or better skills in simulating water and energy fluxes, as verified by in-situ validations. The hyper-resolution simulations offered a detailed and more reasonable description of land surface processes and attained similar spatial patterns and magnitudes with high-resolution land surface data, especially over highly elevated areas. Additionally, this study highlighted the benefits of using mountain radiation theory-based shortwave radiation downscaling models and AutoML-assisted precipitation downscaling models. These findings emphasized the significance of integrating topography-based downscaling methods for hillslope-scale simulations.

Low soil temperature stress is a critical factor affecting the root water uptake (RWU) rate of plants. In current land surface models, the RWU amount is determined by the soil water extracted from different soil layers, which calculates by the relative soil water availability and the root fraction of each layer in the rooting zone. The effect of low soil temperature stress is not considered, which may produce biases in the simulation of transpiration. In this study, with the utilization of the in-situ observation data from three FLUXNET sites, we introduced three functions to represent the low soil temperature stress in the Common Land Model (CoLM) and evaluated their effects on the energy fluxes simulation. Then the three low soil temperature stress functions were also evaluated in the global offline simulations by using the FLUXNET-MTE (multi-tree ensemble) data. Results show that the default CoLM overestimates the latent heat flux but underestimates the sensible heat flux in the local spring and early summer at three study sites. By incorporating the low soil temperature stress function into CoLM, the bias in energy flux simulation is significantly reduced. The global offline simulations indicate that considering the effect of low soil temperature stress can improve the model performance on the simulating of the latent heat flux in those high latitude areas. Therefore, we recommend incorporating the effect of low soil temperature stress into land surface models, which is beneficial to increasing the reliability of the models’ results, especially over the cold regions.