Advancing Precipitation and Transboundary Flood Forecasting in Monsoon Climates

Abstract

About a billion people are directly or indirectly affected by annual monsoon flooding in South and Southeast Asia. Skillful flood forecasting is crucial in this densely populated part of the world, where most of the countries share large international river basins. Flood forecasting is a challenging task for the downstream nations in this region due to lack of upstream in-situ data. Data from global numerical weather prediction (NWP) models are now common to the operational flood forecasting agencies as an alternative to in-situ data. Many of these agencies use the NWP model as a “black box” and the impact of model configurations in operational flood forecasting system has not been extensively studied for monsoon climates. Therefore, it is appropriate to study the performance of this NWP models in monsoon flood forecasting to enhance the current systems. Due to the current lack of structured guidance for operational users of weather and climate data for flood forecasting, performance of the general circulation model (GCM), regional NWP model (Weather Research and Forecasting), and global NWP model (Global Forecasting System) were studied for monsoon regimes. Investigation shows that the GCM are not suitable for operational application at seasonal timescales, where climatology outperforms in persistence based forecasting. Next, regional NWP model (WRF) model configuration was optimized for the monsoon climate before using it for flow forecasting. Through a comprehensive investigation of possible model configurations, three different cloud microphysics and cumulus parameterization schemes were identified as optimal for monsoon climates. Investigations revealed that a generalized forecasting approach is indeed feasible for the operational NWP-based flood forecaster in South and Southeast Asia. Finally, the most user-ready element of this study was derived from a comparison between the regional NWP model (WRF) and global NWP model (GFS) forecasted flow. The results indicate that the improvement due to the use of a regional NWP model like WRF for flow forecasting in large river basins with strong monsoon driven seasonality is marginal compared to that obtained from global NWP-based (GFS) flow forecasting. An easy to apply and computationally efficient bias correction scheme has been developed for operational application of weather forecast forcing from global NWP that can bypass the routine need for dynamic downscaling by regional NWP model. This bias correction scheme further improved skill in flow forecasting thereby making real-world application of global NWP weather forecast forcing computationally efficient in resource-constrained setting of forecasting agencies of South and Southeast Asia.