Kriging-variance based multi-member ensembles of radar-raingauge precipitation estimates: application in Switzerland

CombiPrecip is an operational algorithm of MeteoSwiss that combines in real-time raingauge measurements with radar precipitation estimates across a domain of 710x640km2, covering Switzerland and extending beyond the Swiss borders. The system utilizes a geostatistical approach known as kriging with external drift as an interpolation technique, offering probabilistic outcomes that provide both a mean value and a variance at each interpolated point. The purpose of our study is two-fold: (i) We investigate to what extent the kriging variance of CombiPrecip is a satisfactory measure of uncertainty of the kriging expected value. We answer this question through a probabilistic verification of a seven-year dataset against raingauge measurements. (ii) We present an algorithm which integrates the kriging expected value and variance of the CombiPrecip output with spatially autocorrelated noise fields to generate ensembles of N realistic members. The verification suggests that the probabilistic CombiPrecip output has skill, which remains satisfactory even for high precipitation intensities. The ensembles generated by this method can serve as valuable initial conditions for precipitation nowcasting systems. Moreover, the proposed ensemble-generation technique is not restricted to geostatistics-based applications and can be readily adapted to other approaches that produce probabilistic outputs.