Understanding the Dynamics of Parameter Importance for Global Parameterization of Modeling Semi-Arid Watersheds.

Abstract:

The dynamics of parameter importance in an earth systems modeling framework has been the focus of research in recent years. To investigate the changing aspects of parameter importance, we implemented the variogram analysis of response surfaces to characterize the predictive uncertainty of the KINEROS2 physically-based distributed hydrologic model in the USDA-ARS Walnut Gulch Experimental Watershed (WGEW) and Long-Term Agro-ecosystem Research (LTAR) site. Variogram-based time-variant (throughout the simulation period) and time-aggregate (average of a simulation) parameter importance metrics were assessed to explain model parameter control factors. We explored and quantified the extent of parameter control that varies in space and time due to spatial and temporal distribution of rainfall, such as rainfall intensity, rainfall depth, location of the storm center, and the properties of medium-sized semi-arid watersheds such as the variability of soil and land use. The results showed the importance of the parameters varied considerably depending on the intensity and depth of rainfall, the temporal rainfall distribution patterns, and the location and distance of runoff generating storms relative to the watershed outlet. The application of event-based analysis of parameter importance and calibration models showed a wide range of parameter variability between events that provided good prediction. In these exercises, we developed an approach towards an integrated global K2 parameterization scheme that improved model performance with minimal adjustment of the top two most important parameters identified depending on the rainfall properties (intensity, location, and distributions).