High-resolution land surface modelling over Africa: the role of uncertain soil properties in combination with forcing temporal resolution

Abstract

Land surface modelling runs conducted with the Community Land Model version 5.0 (CLM5) over Africa at 3 km resolution were carried out, and we assessed the impact of different sources of soil information and different upscaling strategies for the soil information, in combination with different atmospheric forcings and different temporal resolutions of those atmospheric forcings. FAO and SoilGrids250m soil information was used. SoilGrids information at 250 m resolution was upscaled to the 3 km grid scale by three different methods: (i) random selection of one of the small SoilGrids250m grid cells contained in the model grid cell, (ii) arithmetic averaging of SoilGrids soil texture values, and (iii) selection of the dominant soil texture. These different soil model inputs were combined with different atmospheric forcing model inputs, which provide inputs at different temporal resolutions: CRUNCEPv7 (6-hourly input resolution), GSWPv3 (3-hourly), and WFDE5 (hourly). We found that varying the atmospheric forcing influenced the states and fluxes simulated by CLM5 much more than changing the soil information. Varying the source of soil texture information (FAO or SoilGrids250m) influences model water balance outputs more than the upscaling methodology of the soil texture maps. However, for a high temporal resolution of atmospheric forcings (WFDE5), the different soil texture upscaling methods result in considerable differences in simulated evapotranspiration (ET), surface runoff, and subsurface runoff at the local and regional scales, which is related to the higher-temporal-resolution representation of rainfall intensity in the model. The upscaling methodology of fine-scale soil texture information influences land surface model simulation results but only when clearly in combination with high-temporal-resolution atmospheric forcings.