The objective of this research was to examine the influence of the spatial variability of precipitation data from different sources on the accuracy of runoff simulation. In this study, interpolated precipitation by external drift kringing and conditional simulated precipitation based on turning band simulation are used to represent the spatial variability of precipitation. Also, the precipitation is averaged over different spatial scales ranging from 4 km2 to 614 km2. The original and averaged precipitation is used as a main forcing input in a conceptual distributed rainfall-runoff model. An automatic calibration method based on the combinatorial optimization algorithm simulated annealing is applied. Also, the aggregated Nash-Sutcliffe coefficients at different time scales are used as objective function in the study. The simulated hydrographs obtained using original and averaged precipitation are analysed through comparisons of the Nash-Sutcliffe coefficient and other goodness-of-fit indexes. By means of them, the influence of spatial variability of precipitation data on the accuracy of simulated runoff is assessed.
Conditional simulation, Distribution modelling, Simulated annealing, Spatial variability
