Coupling Stochastic Interpolation with Budyko Approach for Spatial Prediction of Mean Annual Runoff

Hydro-stochastic interpolation method based on the traditional block-kriging is popularly used to predict mean annual runoff accounting for water balance constraints in basin river system. Nevertheless, this statistic technique does not explicitly interpret the nonstationary variation of the regional average runoff that is affected by spatial variations of climate and landscape over catchments. In this study, the spatial runoff is decomposed into the nonstationary deterministic and stationary stochastic components, interpreted by the Budyko scheme (Fu's equation) and the hydro-stochastic interpolation, respectively. The coupling stochastic interpolation with the Budyko scheme is utilized to interpolate spatial runoff in Huaihe River basin. Results show that the coupling method significantly improves interpolation accuracy. The prediction errors from the coupling method are much less from the respective prediction from the Budyko scheme and the hydro-stochastic interpolation. The cross-validation outcome of the determination efficient from the coupling method is as larger as 0.93, much larger than 0.81 and 0.54 from the Budyko and the hydro-stochastic interpolation, respectively. The prediction from the coupling method best captures regional high and low runoff in space. On contrast, prediction from the Budyko equation and the hydro-stochastic interpolation markedly overestimates low runoff and underestimates high runoff. Thus, coupling hydro-stochastic interpolation with Budyko scheme offers an efficient and accurate way in spatial interpolation of mean annual runoff.