Predictability of stream flow and rainfall based on ENSO for water resources management in Sri Lanka

columbi Summary We investigate the viability of using El Niño–Southern Oscillation (ENSO) and sea surface temperature (SST) data to predict seasonal streamflow for one of the major rivers in Sri Lanka, the Kelani, using correlation analysis, contingency tables, and principal component analysis. The agricultural seasons in Sri Lanka are Yala (April–September) and Maha (October–March). The correlation between the Kelani River streamflow during Yala and ENSO indices (r = 0.41) is significant at 99% level. In addition, the Kelani streamflow during Yala has a correlation with the Central Indian Ocean SST (r = 0.40) that is also significant at the 99% level. The first principal component of the Indo-Pacific Ocean SST is reminiscent of the SST associated with the ENSO mode. A prediction scheme based on this mode for the streamflow during Yala has a skill characterized by a correlation of 0.5 in a cross-validated mode. The prediction of streamflow during Maha is best carried out separately for the two halves of the season. During the El Niño phase, the rainfall during Maha is enhanced during the first half of the season (October–December) and diminished in the second half (January–February). Rainfall rather than streamflow has a better relationship with ENSO from October to December. During the second half of the Maha season, rainfall declines with both warm and cold ENSO phases and any prediction scheme has to take into account this non-linear relationship. Overall, useful skill for seasonal streamflow predictions has been demonstrated for the Yala season and skill for seasonal rainfall predictions for the first and second half of the Maha season has been elucidated. a 2006 Elsevier B.V. All rights reserved.