Assessment of the effects of spatial resolutions on daily water flux simulations

Impacts of spatially distributed precipitation and soil heterogeneity on modeling water fluxes at different spatial resolutions are investigated using the Three-layer Variable Infiltration Capacity (VIC-3L) land surface model at the Blue River watershed in Oklahoma. In this study, hourly grid-based NEXRAD (Next Generation Radar) Stage III radar precipitation data approximately at 4 x 4 km(2) resolution are used to compute daily precipitation at spatial resolutions of 1/32, 1/16, 1/8, 1/4, 1/2 and I degree based on an area weighted average method. Soil parameters at the corresponding six spatial resolutions are derived from the State Soil Geographic (STATSGO) soil data. The forcing data of daily maximum and minimum temperature, wind speed, and vegetation parameters are disaggregated/aggregated directly to finer/coarser spatial resolutions based on the University of Washington (UW) data, which are gridded at 1/8 degree spatial resolution. Our study suggests that a critical spatial resolution for the VIC-3L model may exist for the study watershed. For spatial resolutions finer than the critical resolution, one does not necessarily obtain better model performance in terms of runoff, evapotranspiration, and total zone soil moisture with increasing spatial resolution if the VIC-3L model parameters are calibrated at each spatial resolution. Also, model parameters calibrated at a coarse resolution can be applied to finer resolutions to obtain generally comparable results. However, model parameters calibrated at finer resolutions cannot result in comparable results when applied to resolutions coarser than the identified critical resolution. In addition, while soil moisture of the total zone is more sensitive to the spatial distributions of soil properties, runoff and evaporation are more sensitive to the spatial distribution of daily precipitation at the watershed being studied. (C) 2004 Elsevier B.V. All rights reserved.