Stochastic Models of Spatial and Temporal Distribution of Thunderstorm Rainfall

A simplified stochastic model based on airmass thunderstorm rainfall data from the 58-square-mile Walnut Gulch Experimental Watershed in south eastern Arizona is being developed at the Southwest Watershed Research Center, Tucson, Ariz. Records from the 95 rain gage network on this watershed pro vide valuable information on airmass thunderstorm rainfall in the Southwestern United States. Proba bility distributions are being used to model random variables—number of cells, spatial distribution of the cells, and cell center depths—of thunderstorms in a summer rainy season. A computer program produces synthetic thunderstorm rainfall based on these distributional assumptions. The synthetic data are compared, with respect to storm center depths and isohyetal map characteristics, with data from the dense rain gage network on Walnut Gulch. The daily and hourly chances of occurrence of seasonal airmass thunderstorm rainfall are modeled. Efforts are being made to model the temporal dis tribution of rainfall from individual cells within the airmass thunderstorm. Finally, the question of model transferability to other regions and locations is tied to defining regional meteorology and local topography.