Improving the flood forecasting capability of the Xinanjiang model for small- and medium-sized ungauged catchments in South China

Prediction in ungauged basins (PUB) is as crucial as it is challenging. Thus far, there have been abundant regionalization studies on PUB, whereas "regionalization" is the main research method. In order to estimate Xinanjiang model parameters in ungauged areas and improve the accuracy of flood simulation for small and medium-sized ungauged catchments, the Xinanjiang model was applied on 33 mountainous small- and medium-sized catchments in south China. This study investigated the relative benefits of traditional regionalization methods (physical similarity and parameter regression) and physically consistent parameter estimation method. The latter can directly estimate three sensitive parameters of the Xinanjiang model without the need of regionalization. In addition, the effect of the number of donor catchments was addressed. The results show that the prediction accuracy of the traditional regionalization methods did not obtain satisfactory prediction results. However, the integrated schemes, which combine the regionalization methods with physically consistent methods, performed considerably better than the traditional regionalization methods, indicating that directly performing a parameter estimation from underlying surface data of ungauged catchments can improve the transferability of the Xinanjiang model in these catchments. On the other hand, the best accuracy was obtained when the number of donor catchments was equal to five in the integrated schemes. The integrated parameter estimation schemes proposed in this study support more effective hourly flood event simulation for small- and medium-sized ungauged catchments in southern China.