Scalability of grid- and subbasin-based land surface modeling approaches for hydrologic simulations

This paper investigates the relative merits of grid- and subbasin-based land surface modeling approaches for hydrologic simulations, with a focus on their scalability (i.e., ability to perform consistently across spatial resolutions) in simulating runoff generation. Simulations are produced by the grid-and subbasin-based Community Land Model at 0.125 degrees, 0.25 degrees, 0.5 degrees, and 1 degrees spatial resolutions over the U.S. Pacific Northwest. Using the 0.125 degrees simulation as the reference solution, statistical metrics are calculated by comparing simulations at 0.25 degrees, 0.5 degrees, and 1 degrees resolutions with the 0.125 degrees simulation for each approach. Statistical significance test results suggest significant scalability advantage for the subbasin-based approach compared to the grid-based approach. Basin level annual average relative errors of surface runoff at 0.25 degrees, 0.5 degrees, and 1 degrees resolutions compared to the 0.125 degrees simulation are 3%, 4%, and 6% for the subbasin-based configuration and 4%, 7%, and 11% for the grid-based configuration, respectively. The scalability advantages are more pronounced during winter/spring and over mountainous regions. The source of runoff scalability is found to be related to the scalability of major meteorological and land surface parameters of runoff generation. More specifically, the subbasin-based approach is more consistent across spatial scales than the grid-based approach in snowfall/rainfall partitioning because of scalability related to air temperature and surface elevation. Scalability of a topographic parameter used in runoff parameterization also contributes to improved scalability of the rain-driven saturated surface runoff component, particularly during winter. Hence, this study demonstrates the importance of spatial structure for multiscale modeling of hydrological processes.