Search

This paper investigates whether the calibrated parameter values for two rainfall–runoff models based on historical observed data can be used to reliably predict runoff responses to changes in future climate inputs. Analyses are carried out using historical monthly climate and runoff data from 27 catchments in West and Central Africa. Better performances are obtained when the models are calibrated on a dry period to predict runoff in a wet period. Similarly, it is easier for the models calibrated on a period of large hydroclimatic variability to predict runoff in a period of small hydroclimatic variability than the contrary. This paper tests also the assumption that transferability of rainfall–runoff models, variable hydroclimatic conditions, and local sensitivity of the model parameters can be linked. The inconclusive results still show variability according to the chosen efficiency criteria and selected periods. All these results indicate that the transfer of model parameters from one period to another can introduce a significant error in runoff simulations, which means increased uncertainty in the assessment of impact of climate and environmental changes, an essential question for West and Central Africa.Editeur M.C. Acreman; Editeur associé Gil Mahé [ABSTRACT FROM PUBLISHER]