| description abstract | Statistical relationships between topography and the spatial distribution of mean annual precipitation are developed for ten distinct mountainous regions. These relationships are derived through linear bivariate and multivariate analyses, using six topographic variables as predictors of precipitation. These predictors are elevation, slope, orientation, exposure, the product (or interaction) of slope and orientation, and the product of elevation and exposure. The two interactive terms are the best overall bivariate predictors of mean annual precipitation, whereas orientation and exposure are the strongest noninteractive bivariate predictors. The regression equations in many of the climatically similar regions tend to have similar slope coefficients and similar y-intercept values, indicating that local climatic conditions strongly influence the relationship between topography and the spatial distribution of precipitation. In contrast, the regression equations for the tropical and extratropical regions exhibit distinctly different slope coefficients and y-intercept values, indicating that topography influences the spatial distribution of precipitation differently in convective versus nonconvective environments. The multivariate equations contain between one and three significant topographic predictors. The best overall predictors in these models are exposure and the interaction of elevation and exposure, indicating that exposure to the prevailing wind is perhaps the single most important feature relating topography to the spatial distribution of precipitation in the mountainous regimes studied. The strongest (weakest) multivariate relationships between topography and precipitation are found in the four middle- and high-latitude west coast regions (in the tropical regions), where more than 70% (less than 50%) of the spatial variability of mean annual precipitation is explained. These results suggest that in certain regions, one can estimate the spatial distribution of mean annual precipitation from a limited network of raingauges using topographically based regression equations. | |
