Spatial Interpolation of Sediment Yield Estimated from Reservoir Siltation Data of India

Abstract

Estimation of design sediment yield (SY) on a long-term basis is necessary to allocate adequate reservoir dead storage space. Incompetence of the distributed SY models with the input data (of spatial and temporal scale) leads to an error in model results. Therefore, prudent management of the calibration and validation process of SY predicting model is essential. The absence of the observed SY data brings indeterminacy in the validation process. However, a reference value for validation can be determined with spatial interpolation of the observed regional SY. In the present study, 161 spatially distributed Indian reservoirs were analyzed. Considering SY computed from these reservoirs as sampling values, inverse distance weighting and various Kriging interpolation methods were used to generate 25 interpolated datasets. Optimization of interpolation parameters is carried out by an exhaustive cross-validation process, and the best-interpolated surface is identified. Comparing the observed and predicted SY, the coefficient of determination is found to be 0.78, with the index of agreement being 0.88. This obtained surface model was utilized to generate a sediment yield contour map for India.