Analysis of conceptual rainfall–runoff modelling performance in different climates

With the objective of studying conceptual rainfall–runoff modelling performance in different climates, the HBV-96 model was applied on four catchments located in Europe, Africa and South America. Manual, automatic and Monte Carlo techniques were used for model calibration and parameter analyses. It was found that the magnitude of the water balance components had a significant influence on model performance. Performance decreased and demands of calibration period length increased with increased catchment dryness primarily because of a neater water balance and higher climatic variability in drier areas. A large degree of equifinality was discovered in all catchments where different calibration methods yielded equally good results but with different parameter combinations. Thus, it may be impossible to know if an optimum parameter set exists and to relate parameter values to physical properties of the catchment. On the other hand the validation results indicated that it might not matter if parameter values were not unique when studying runoff solely, provided the model application is within the same regime of flows.