Predicting the streamflow regimes using climate dynamics is important in water resource management. However, in Indonesia, there are few studies targeting climate indices and streamflow. A previous study found difficulty in developing a statistical prediction model for this relationship due to its non-linear nature. This study attempted to address that gap by applying multiple regression (MR) models using second- and third-order polynomial functions to show the non-linear relationship between climate and flow regime indices. First, a correlation analysis was performed to check the variable relationships. There was a good and significant correlation of El Niño Southern Oscillation (ENSO) with the flow regime indices. Secondly, MR models were developed with the same-time variables. The developed model showed that the Indian Ocean Dipole (IOD) had the effect of strongly increasing the high flow in La Niña phases. Finally, time-lagged MRs were developed aiming at forecasting. Lagged MR models with six-month leading climate indices demonstrated a relatively good correlation with the observed data (mostly R > 0.700) with moderate accuracy (root mean square error = 44–51%). It suggests that the forecasting of flow regime may be possible using ENSO and IOD indices.
While many studies on the relationship between climate modes and rainfall in Indonesia already exist, studies targeting climate modes’ relationship to streamflow remain rare. This study applied multiple regression (MR) models with polynomial functions to show the teleconnection from the two prominent climate modes—El Niño–Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD)—to streamflow regimes in eight rivers in Java, Indonesia. Our MR models using data from 1970 to 2018 successfully show that the September–November (SON) season provides the best predictability of the streamflow regimes. It is also found that the predictability in 1970–1989 was better than that in 1999–2018. This suggests that the relationships between the climate modes and streamflow in Java were changed over periods, which is suspected due to the river basin development. Hence, we found no clear spatial distribution patterns of the predictability, suggesting that the effect of ENSO and IOD are similar for the eight rivers. Additionally, the predictability of the high flow index has been found higher than the low flow index. Having elucidated the flow regimes’ predictability by spatiotemporal analysis, this study gives new insight into the teleconnection of ENSO and IOD to the Indonesian streamflow.
1 January 2011 in Journal of Agricultural Meteorology