The role of probabilistic precipitation forecasts in hydrologic predictability

Abstract

Accurate streamflow forecasts enable the appropriate management of water resources. Although there is a general consensus that climate information can enhance hydrological predictability, this might not be the case if the accuracy of the given climate information is unreliable. Hence, this study has developed a modeling framework to estimate the role of climate information in forecasting accurate streamflow. Ensemble streamflow prediction (ESP) technology was adopted as a dynamic hydrologic forecast method to 35 watersheds in South Korea. The probabilistic precipitation forecast (PPF), issued by the Korea Meteorological Administration, was used as climate information for updating the probabilities of climate scenarios. First, we found that the current PPF is not accurate enough for significantly enhancing the streamflow forecasting accuracy. Subsequently, multiple sets of PPF were synthetically generated to evaluate the role of climate information. Given the perfect categorical climate forecasts, we found that there is much potential for the enhancement of streamflow forecast skill especially in the seasons that exhibit greater streamflow variability. However, there is less potential for increasing the streamflow forecasting skill under below-normal conditions. The proposed modeling framework is capable of quantifying the magnitude of potential improvement in hydrological predictability under the assumption that better climate information will be available in the future. We expect that this modeling framework can be effectively applied to other regions across a wide range of climate regimes.