Spatial estimation of daily precipitation in regions with complex relief and scarce data using terrain orientation

Precipitation is the most relevant variable in the hydrological cycle which drives continental hydrologic processes. Its spatial occurrence and behavior are complex and its daily estimation is hard in poorly gauged regions where the topography is highly irregular. Several interpolation methods are available for this purpose, but their performance is quite uncertain. This study develops a spatial interpolation method for daily precipitation that considers both spatial discontinuities and the influence of topography. The method first identifies the precipitation occurrence in each grid-cell as a function of measurements in surrounding rain gauges, and then uses daily elevation vs. precipitation linear regressions throughout the grid-cells where precipitation occurrence is identified. These regressions are classified according to the terrain orientation with respect to the prevailing wind direction. The method was evaluated using categorical statistics that quantify the skill to identify the precipitation occurrence/non-occurrence, and goodness-of-fit statistics to evaluate the error and efficiency. The methodology was compared against inverse distance weighted and simple regression methods, which were implemented considering both continuous and discontinuous precipitation fields. The new method better simulates the occurrence of precipitation, whereas traditional methods applied without considering the spatial discontinuity of precipitation tend to overestimate the frequency of the rainfall events, and thus the magnitude of precipitation at the basin scale. When spatial discontinuity is considered, traditional methods improve their performance and are comparable to the proposed method. Overall, the new method increases the number of days in which elevation vs. precipitation linear regression can be used, thus improving the spatial representation of precipitation in areas with complex relief.