Abstract
This study aims to develop a method for forecasting the water level of the Kahayan River in Indonesia. In recent years, peatland in that basin has been drying as a result of agricultural development, and such drying has led to more frequent large-scale wildfires there. By incorporating the sea surface temperature, which is affected by the El Nino event, into a Nearest-Neighbor Method (NNM), the water level forecasting accuracy was improved even in case that the lead time of prediction was extended. In addition, highly accurate water level forecasts were obtained by using predicted rainfall calculated by NNM and a tank model that can reproduce the hydrological cycle in the basin. Accordingly, a method for forecasting the water level even when data are missing was proposed. The results promise to be useful for basin management to prevent peatland wildfires.
