Distance in spatial interpolation of daily rain gauge data

Abstract

Spatial interpolation of rain gauge data is impor-tant in forcing of hydrological simulations or evaluation ofweather predictions, for example. This paper investigatesthe application of statistical distance, like one minus com-mon variance of observation time series, between data sitesinstead of geographical distance in interpolation. Here, asa typical representative of interpolation methods the inversedistance weighting interpolation is applied and the test datais daily precipitation observed in Austria. Choosing statis-tical distance instead of geographical distance in interpola-tion of available coarse network observations to sites of adenser network, which is not reporting for the interpolationdate, yields more robust interpolation results. The most dis-tinct performance enhancement is in or close to mountainousterrain. Therefore, application of statistical distance in theinverse distance weighting interpolation or in similar meth-ods can parsimoniously densify the currently available obser-vation network. Additionally, the success further motivatessearch for conceptual rain–orography interaction models ascomponents of spatial rain interpolation algorithms in moun-tainous terrain.