Regional calibration of catchment models:: Potential for ungauged catchments

[1] We present a novel iterative regional calibration (IRC) method in which the model parameters of a number of catchments are calibrated simultaneously. The method exploits the spatial correlations of the parameters to condition their a priori distribution for each catchment. We use a sample of 320 catchments in Austria over a period of 22 years to test the method. The results indicate that the IRC method allows us to reliably calibrate the conceptual hydrologic model used here. The novel method reduces the uncertainty of most parameters as compared to local calibration. This is demonstrated by more consistent model parameters in two independent calibration periods and by an analysis of their spatial variability. Jackknife cross validation indicates that the IRC method tends to improve runoff simulation performance for ungauged catchments as compared to traditional regionalization, although the gain is small in absolute terms.