A comparison of regionalisation methods for catchment model parameters

摘要

In this study we examine the relative performance of a range of methods fortransposing catchment model parameters to ungauged catchments. We calibrate11 parameters of a semi-distributed conceptual rainfall-runoff model todaily runoff and snow cover data of 320 Austrian catchments in the period1987-1997 and verify the model for the period 1976-1986. We evaluate thepredictive accuracy of the regionalisation methods by jack-knifecross-validation against daily runoff and snow cover data. The resultsindicate that two methods perform best. The first is a kriging approachwhere the model parameters are regionalised independently from each otherbased on their spatial correlation. The second is a similarity approachwhere the complete set of model parameters is transposed from a donorcatchment that is most similar in terms of its physiographic attributes(mean catchment elevation, stream network density, lake index, arealproportion of porous aquifers, land use, soils and geology). For thecalibration period, the median Nash-Sutcliffe model efficiency ME of dailyrunoff is 0.67 for both methods as compared to ME=0.72 for the at-sitesimulations. For the verification period, the corresponding efficiencies are0.62 and 0.66. All regionalisation methods perform similar in terms ofsimulating snow cover.