Impacts of spatial resolutions on water budget simulated by the VIC-3L hydrologically-based land surface model

Impacts of spatially-distributed precipitation and soil heterogeneity on water fluxes at different spatial resolutions are explored using the VIC-3L (Three-layer Variable Infiltration Capacity) land surface model at the Blue River basin, OK. In the study, hourly grid-based NEXRAD stage ØV radar precipitation data (about 4x4 km2) are used to obtain daily precipitation at six different resolutions (i.e. 1/32nd, 1/16th, 1/8th, 1/4th, 1/2nd and 1 degree) based on an area weighted average method. Soil parameters at the corresponding six resolutions are derived from the STATSGO (State Soil Geographic) data set. The forcing data of daily maximum and minimum temperature and wind speed needed by the VIC model, and the vegetation parameters are disaggregated/aggregated directly to higher/lower resolutions based on the University of Washington_fs data gridded at the 1/8th degree resolution. Our study suggests that there seems to exist a critical spatial resolution for VIC-3L at which higher spatial resolution does not necessarily result in better model performance if the model parameters are calibrated separately for each resolution. Also, model parameters calibrated at a coarse resolution can be applied to the higher resolutions with comparable results. However, model parameters calibrated at the higher resolutions cannot yield results comparable to applying the same parameters at the lower resolutions. In addition, compared to evaporation, soil moisture is more sensitive to soil properties, while evaporation is more sensitive to precipitation.