Estimation of catchment rainfall uncertainty and its influence on runoff prediction

Abstract

Interpolation of spatially varying point precipitation depths introduces uncertainties in the estimated mean areal precipitation (MAP). This paper describes a geostatistical approach - the Kriging method - to calculate the daily MAP on real-time basis. The procedure provides a linear unbiased estimate with minimum estimation variance. The structural analysis of the random precipitation field is automatized by relating the time-varying semivariogram model to the sample variance. This is illustrated on data from a Danish IHD catchment. The conceptual rainfall-runoff model NAM incorporated into a Kalman-filter algorithm is applied to investigate the effects of uncertainties in MAP on the runoff predictions. Measurement and processing errors are not included in the investigation. -Authors