Importance analysis of local and global climate inputs for basin-scale streamflow prediction

摘要: [1] Basin-scale streamflow is influenced by numerous local and global climate inputs. In this paper, genetic programming (GP) combined with “importance analysis” to identify the important inputs meteorological variables needed for prediction of weekly at basin scale. The analysis carried out Mahanadi River in India using inputs, namely, El Nino–Southern Oscillation (ENSO) index equatorial Indian Ocean (EQUINOO) index; including outgoing longwave radiation (OLR), total precipitable water (TPW), temperature anomaly (TA), pressure (PA); information from previous time steps. rainfall over intentionally not utilized so that procedure may be applicable basins little or no rain gauge achieve a longer lead time. Birnbaum importance measure used assess each input. Results study show relative individual input lags. It observed among various OLR PA are more than TA TPW. Among large-scale circulation indices, ENSO 5th 7th week, whereas EQUINOO 3rd 6th week. On basis their measures, 15 indices were selected initial group 30 indices. GP-derived forecasting models could predict good accuracy (correlation coefficient r = 0.821) such complex system.