Comparison of regionalization approaches based on regression and similarity for predictions in ungauged catchments under multiple hydro-climatic conditions

Streamflow predictions in ungauged catchments are a challenge for hydrologists around the world. Using the Variable Infiltration Capacity model, the comparison of two regionalization approaches - one based on regression and the other on similarity - was investigated in 55 catchments in China under multiple hydro-climatic conditions. The regression-based approach was based on multiple regression equations using the calibrated model parameters as the dependent variables and catchment physical characteristics as the independent variables. The similarity-based approach considered the weights of five donor catchments, according to a combination of physical similarity and spatial proximity. The comparison of the two regionalization approaches indicated that the similarity-based approach was more accurate than regression-based approach, both for humid regions and arid regions; with greater aridity, the similarity-based approach became increasingly more accurate than the regression-based approach. The uncertainty in the comparison of the different regionalization approaches should be investigated by more catchments in future studies. (C) 2012 Elsevier B.V. All rights reserved.