Improved Method for Multi-GCM Ensemble Averaging of Downscaled CMIP5 Precipitation for Capturing the Observed Precipitation Series

Prediction of precipitation has always been a challenge for hydrologists due to its erratic nature. The most recent advances in this discipline are General Circulation models (GCMs), which are employed for prediction and projection of precipitation. A Number of agencies have developed their own GCMs following the norms provided by Intergovernmental Panel on Climate Change, yet every GCM cannot be stated as ideal for a particular location. In this paper, an attempt is made to capture the pattern in observed precipitation time series by employing downscaled multi-GCM precipitation ensemble through averaging. Prior to averaging, ranking for best fitting GCMs is done on the basis of certain statistical parameters in order to determine the GCMs with best precipitation capturing capability. A number of combinations of highest ranked GCMs are made as per the ranking obtained from statistical analysis. The GCMs in this combination are averaged using a regression method for simulating the observed precipitation. The Method of M5 pruned model rules (using smoothed linear models) is practiced as a regression purpose for averaging of Multi-GCM Ensemble. A total of 39 GCMs are used from CMIP5 archive. Above proposed methodology is illustrated for predicting precipitation over Yamuna basin.