Influence of rainfall observation network on model calibration and application

摘要

The objective in this study is to investigate the influence of the spatialresolution of the rainfall input on the model calibration and application.The analysis is carried out by varying the distribution of the raingaugenetwork. A meso-scale catchment located in southwest Germany has beenselected for this study. First, the semi-distributed HBV model is calibratedwith the precipitation interpolated from the available observed rainfall ofthe different raingauge networks. An automatic calibration method based onthe combinatorial optimization algorithm simulated annealing is applied. Theperformance of the hydrological model is analyzed as a function of theraingauge density. Secondly, the calibrated model is validated usinginterpolated precipitation from the same raingauge density used for thecalibration as well as interpolated precipitation based on networks ofreduced and increased raingauge density. Lastly, the effect of missingrainfall data is investigated by using a multiple linear regression approachfor filling in the missing measurements. The model, calibrated with thecomplete set of observed data, is then run in the validation period usingthe above described precipitation field. The simulated hydrographs obtainedin the above described three sets of experiments are analyzed through thecomparisons of the computed Nash-Sutcliffe coefficient and severalgoodness-of-fit indexes. The results show that the model using differentraingauge networks might need re-calibration of the model parameters,specifically model calibrated on relatively sparse precipitation informationmight perform well on dense precipitation information while model calibratedon dense precipitation information fails on sparse precipitationinformation. Also, the model calibrated with the complete set of observedprecipitation and run with incomplete observed data associated with the dataestimated using multiple linear regressions, at the locations treated asmissing measurements, performs well.