Assessment of an analogue downscaling method for modelling climate change impacts on runoff

SummaryDownscaling is a technique commonly used to transform the outputs of a low resolution climate model to a finer scale suitable for regional hydrological modelling. The technique is an important component in studies investigating the impact of climate change on runoff. This study assesses a simple analogue downscaling method – the statistically based method of Timbal and McAvaney (2001) to provide daily rainfall series for modelling runoff across southeast Australia. Using 10 global climate models (GCMs), the historical and future rainfall and runoff generated by this method were compared with the results from an empirical scaling method. We find that the analogue method can be used with hydrological models to simulate long-term average runoff over large regions reasonably well provided a suitable ‘inflation factor’ is applied. Without the correction, biases are too large for direct input to hydrological models. Even with appropriate inflation, the daily rainfall distribution from the analogue method will not be the same as the observed distribution, and this may lead to unreliable simulations of daily runoff characteristics. Comparing the analogue method and the empirical scaling method on the basis of change in future runoff, averaged across the entire study region, the results are generally similar with large majority of the GCMs showing a decline in future runoff. In the cases when there are differences, the additional climatic response produced by the analogue method appears to be a consequence of using atmospheric moisture variable as a predictor in certain regions. The range in the future rainfall and runoff projections from the analogue method is smaller than that from the empirical scaling. This is because compared to the empirical scaling, which uses rainfall derived from as many as 10 different parameterisations in the GCMs, the analogue method uses a single relationship between synoptic atmospheric fields and rainfall for all the GCMs. This, and the fact that the analogue method models future changes at the relevant catchment scale and captures changes to a larger range of rainfall characteristics, are the advantages of the analogue method. Nevertheless, more research is required to improve daily rainfall series for direct input into hydrological models to fully realise the potential of analogue downscaling for hydrological applications.

► We assessed an analogue downscaling method in modelling climate impacts on runoff.
► The historical runoff from this method is reasonably well produced after inflation.
► The method can reduce range in the future rainfall and runoff projections from GCMs.
► Regionalisation, choice of predictors, and inflation may introduce uncertainty.