PNNL

Abstract

In this study, we used downscaled high resolution climate simulations to provide inputs to the physics-based Distributed Hydrology Soil Vegetation Model (DHSVM), which accounts for the combined effect of rainfall and snowmelt processes, to determine the spatially distributed available water for runoff (AWR). Model outputs were extracted for two different mountainous field sites in Colorado and California. IDF curves for precipitation and AWR were generated and compared at each numerical grid. Quantitative evaluation of trending and stationarity tests were conducted to identify (quasi-)stationary time periods for reliable IDF analysis, which ensures that the IDFs are developed without violating the assumption of stationarity. By dividing the data into 30-year time windows, both AWR and precipitation annual maxima series are quasi-stationary, although in some local areas they exhibit relatively strong temporal trends. Impacts of snowmelt are found to be spatially variable due to spatial heterogeneity such as topography and vegetation cover. Geostatistical analyses were used to understand the spatial patterns of extreme events, and AWR and precipitation intensity were then spatially mapped for events with various durations and frequencies.