Revisiting and Improving Soil Surface Resistance Schemes for Land Surface Modeling

Abstract

Accurate representation of soil evaporation is critical for understanding land-atmosphere interactions and improving climate simulations, where soil surface resistance plays a key regulatory role. However, existing schemes differ substantially and lack systematic evaluation. We compared five typical schemes and found large discrepancies in soil surface resistance values and site-level simulations. Under intermediate soil wetness, different schemes showed over tenfold differences and had differing sensitivity parameters. Incorporating these schemes into the Common Land Model (CoLM) markedly altered soil wetness, evaporation, energy fluxes and ecosystem productivity, with mean soil wetness increasing by approximately 5%–35% and evaporation decreasing by approximately 9%–73% in summer. Using FLUXNET data and machine learning (XGBoost), we identified sand volume fraction as a key driver of β (a factor reflecting evaporation suppression calculated from soil resistance, with smaller values indicating stronger suppression). Building on Sellers et al. (1992, https://doi.org/10.1029/92jd01096) (S92), widely adopted for its simplicity and effectiveness in simulating latent heat flux and soil moisture, we incorporated this parameter to develop S92_sand, which better reproduced observation-derived β than the typical schemes. When implemented in CoLM, S92_sand reduced mean absolute error and root mean square error (RMSE) of soil wetness by approximately 10% at evaporation-dominant sites compared to S92. Regional assessments across China using GLEAM v4.2a demonstrated its superiority over the original S92 scheme, mitigating systematic overestimation of soil evaporation and reducing RMSE across 90% of the region, with improvements up to 1 mm day⁻¹ over the Qinghai-Tibet Plateau and the North China Plain. The simulations of soil wetness and evapotranspiration also showed noticeable regional improvements. This study provides new insights into soil surface resistance and advances land surface modeling of soil evaporation.