The average validation and extrapolation error statistics of the combined GPM framework are given in Table 7. Similar to ANN, the performance of GPM in the extrapolation also turns out to be quite satisfactory. Both the surrogates show comparable performance as captured by the performance metrics. They both have very low values of NRMSE, NQ95, and NMBE, and very high values of NSE, R, and threshold statistics. Performance of GPM is relatively better, particularly in the extrapolation range.