The impact of uncertainty between PET methods on projections of river flow under climate change in the upper Waikaia catchment, New Zealand.

Appropriate planning and management strategies need to occur to allow countries and regions to deal with the river response to climate change. Within climate change impact assessment studies, there is a high degree of uncertainty at a variety of different scales. These sources of uncertainty range from the methodology of generating climatic variables to the different GCMs used. Potential evapotranspiration (PET) has been overlooked as a source for uncertainty in past studies. This study will focus on the uncertainty produced by PET methods in climate change impact assessment studies of water resources, using the Waikaia catchment in southern New Zealand as a case study. The aim was achieved by using pattern scaled outputs from five GCMs simulating the climatic response to a 2 ºC rise in global mean temperature. The GCM outputs were run through a semi-distributed hydrological model (HBV-Light) that simulated the Waikaia catchment. The output from each GCM was used as input for six different PET methods (Penman, Granger, Hamon, Hargreaves-Samani, Priestley Taylor and Jensen Haise). Under current climate conditions, there is a clear difference in the calculated PET values between the six different PET methods. Under the 2 ºC climate change scenario, all the temperature and radiation based PET methods project an increase in annual PET by about 7-15%. The physical methods response was between -4% and 13% depending on the GCM used. The runoff change is most sensitive to different PET methods in the summer months, in the winter months the runoff sensitivity is low. Overall, there is a low runoff sensitivity runoff to the PET methods for the modelled Waikaia catchment. There is a small simulated increase in runoff over summer, increases by 10% over winter and a reduction in runoff between 10-30% depending on the GCM. The changes in simulated runoff are dominated by the precipitation signal under climate change. The current study is the first investigation carried out in New Zealand where PET uncertainty impacts under climate change have been examined in detail. Contrary to studies elsewhere, the results of this study suggest that for the modelled Waikaia catchment, the selected PET method does not have a large impact on the runoff response to climate change. The low sensitivity to PET could be due to the high rainfall and cool temperatures in the Waikaia, or the calibrated hydrological model of the Waikaia may have a low sensitivity to increasing PET. Further work is needed to confirm if hydrological models are insensitive to PET in a New Zealand context