High-Resolution Soil Data in Land Surface Modelling: Reducing Uncertainties in Heat Balance Simulations for Siberian Ecosystems

Abstract This study highlights the critical role of accurately representing soil hydrophysical coefficients (SHC) in land surface models like the Russian TerM model. It shows that using constant SHC values by soil type—a method suitable for low-resolution models (100–500 km)—leads to large errors when applied to high-resolution modelling (5–50 km). In complex landscapes, such as the Western Siberian peatlands, this approach results in summer temperature differences of up to 9.5°C and soil moisture differences exceeding 60%. Model experiments confirm that both spatial resolution and the method of aggregating SHC data significantly impact accuracy. Aggregating high-resolution SHC data (~ 1 km to ~ 10 km) increased differences, while different averaging methods (arithmetic, geometric, harmonic) produced varying results. The best outcomes were achieved using a combined approach, matching the method to the physical nature of each SHC parameter. The grid resolution of the TerM model itself also affected outputs: coarser grids underestimated soil temperature and latent heat fluxes. Model validation at the Bakchar station showed that correct SHC input reduced temperature errors to under 1.7°C, while constant SHC values doubled the error. The study advocates adopting global high-resolution SHC datasets and adaptive aggregation algorithms to improve model accuracy, especially for climate simulations.