Using a GIS to Estimate the Spatial Variation of Precipitation Due to Topography

Statistical methods can be used to interpolate point precipitation data over space by considering precipitation as the sum of a deterministic component that varies smoothly over space plus a randomly varying residual component. This study estimates the spatial variation of the deterministic component of precipitation due to topography. A method has been developed to define geographical areas in such a way that precipitation can be expressed as a simple function of location within each area. The method consists of using a GIS to (1) define topographic ridge lines based on a spatial averaging scale A, (2) construct idealized sloping areas enclosing each ridge line in a corridor of half-width L, and (3) divide the sloping areas into polygons based on aspect. Precipitation within a polygon is expressed as a function of distance from the nearest ridge line. Five sets of polygons, each set covering the contiguous US, were defined using various combinations of scale parameters A and L. The mean square error in fitting precipitation to location within the polygons was used to select the set of polygons that best represents the spatial variation of precipitation due to topography.