A set of regression equations relating storm rainfall depth to watershed topography and storm type was derived for the high-density precipitation network at Coweeta Hydrologic Laboratory. The most general equation predicted storm amounts for an independent test group of gages with an average error of 0.38 cm (0.15 inches). The dependent variable was the ratio of the rainfall at each gage site to the rainfall at a base gage. Predictive variables were topographic slope, aspect, ground elevation at the gage site, and smoothed elevation. The smoothed elevation, which is the elevation the gage would assume if it were on a smooth plane representing the general topography of the terrain, appeared in more equations than any other variable. One equation was calculated for each of six identified storm types, and one equation was calculated with all storms considered together. Overall, the equations which considered storm type were not better predictors of site rainfall than the equation which did not consider storm type. Predictions of storm amounts were closest to measured amounts for storms where the low-pressure center passed east of Coweeta, while the predictions for the air-mass or thunderstorm type had the greatest errors. The prediction errors of the equations for warm-, cold-, and stationary-front storm types were intermediate. The small number of tropical storms limited development and testing of equations for that type.INTRODUCTION Point values of precipitation in mountainous terrain have great spatial variability due to orographic effects. Much of the early research into the effects of orography on precipitation considered annual rainfall and the importance of elevation on the quantity of annual rainfall [Lee, 1911;Henry, 1919;Barrows, 1933;Price and Evans, 1937]. Later investigators looked at seasonal aspects of orographic rainfall and the influence of other physiographic variables on rainfall in the mountains [Donley and Mitchell, \939;Spreen, \941\Burns, 1953], and the importance of synoptic weather type on determining orographic effects [Williams and Peck, 1962]. Today, research into orographic rainfall includes physical modeling [Struzer, 1972;Collier, 1975; Cotton, 1976], statistical modeling, and observation of the distribution of storm rainfall in mountainous terrain [Mervaetal., 1976;Storebo, 1976;Browning et al, 1975].Dense raingage networks, needed to obtain a representative measure of the areal rainfall distribution in mountainous terrain, are expensive to install and maintain. To estimate rainfall at any point on a watershed given only a single measurement site, we propose to use a rainfall ratio (rainfall at any site on a watershed divided by rainfall at a base measurement station). In application, the ratio would be multiplied by the base station rainfall to obtain a point estimate of rainfall. The primary objective of the study was to develop a method of estimating mean rainfall ratios from independent variables describing the physiography of the ungaged sites. Multiple linear regression techniq...