Estimation of Prediction Interval in ANN-based GCM Downscaling

The General Circulation Model (GCM) is one of the most widely used tools for weather forecasting and assessing the impact of the climate change on the weather parameters. In this study, statistical downscaling of GCMs were applied via Artificial neural network (ANN). As the classic ANN method (point prediction), conveys no information about the accuracy of the prediction, so prediction intervals were used for quantifying the accuracy of downscaling via ANN. For calculating the prediction intervals, novel Lower Upper bound estimation method was used, in which an ANN with two outputs was used to estimate the bounds. Also, Bootstrap method as a classic technique for assessing uncertainty of ANN was used to further examine the proposed LUBE method. In this way, the accuracy of PIs was quantified by coverage and width criteria. Three GCMs, Can-ESM2, BNU-ESM, INM-CM4 and ensemble-GCM (ensemble of them) were used in four grid points around each of station for evaluating ANN-based downscaling of precipitation and temperature parameters. Comparison between the results of two methods indicated that LUBE method could lead to more reliable results than the Bootstrap method. PIs width and coverage probability were 10% to 40% lower and 2% to 10% higher than the Bootstrap method for different GCMs, respectively. Ensemble-GCM led to more accurate results so that computed PIs width and coverage probability were 10% to 60% lower and 2% to 20% higher than those for the single GCMs.