The impact of different evapotranspiration models in rainfall runoff modelling using HBV- light

Abstract: The choice of which potential evapotranspiration (PET) model to use when estimating streamflow using a rainfall-runoff model have been the topic of many studies. The aim with this thesis was to assess the robustness of six different PET models using HBV-light as rainfall-runoff model over three catchments in Sweden. The robustness was evaluated by using a differential split sample test (DSST) based on four climatic conditions with regard of temperature and precipitation. Data from a period of 24 years (1997-2020) was used in order to get a wide range of climatic conditions. The calibration was based on the objective function Kling-Gupta Efficiency (KGE), while the validation of the model was evaluated based on the Nash-Sutcliffe efficiency (NSE) and the volume error (VE). These objective functions are commonly used when evaluating streamflow and are know to provide good estimations of the model performance. The result showed a large difference in efficiency between each PET model and between each catchment (ranging from a KGE of 0.85-0.54 and a NSE of 0.68-0.07), proving that the optimal choice of PET model may be site specific. The Hargreaves-Samani and the Jensen-Haise model were the two PET models which showed an acceptable performance for both calibration and validation over all catchments. It could be noted that calibration and validation on similar climatic conditions would provide models with higher efficiency and that the model parameters are climate dependant.