Flash flood forecasting in West Africa: Evaluating the performance of regional and global numerical weather prediction models

Abstract

The combined effect of population growth, urbanization, and intensification of extremes due to climate change is expected to increase the frequency as well as the magnitude of flash flood impacts in West Africa. It is therefore imperative to develop accurate forecasting procedures that can inform decision making by civil protection and other relevant authorities to mitigate the risk and reduce losses in terms of lives and property. Flash flood events are particularly difficult to predict because they develop at small space and time scales. High-resolution distributed hydrologic models can be used to adequately predict catchment hydrologic response during flash floods, but the accuracy of their forecasts is largely dependent on the quality of precipitation input. In data scarce regions such as West Africa, where a network of ground weather radar is absent, quantitative precipitation forecasts (QPFs) from numerical weather prediction (NWP) models provide an important source of information for flash flood forecasting. QPFs are available from global forecasting systems and regional NWP models at varying spatial resolution, lead time, and quality. To evaluate the operational use of different QPFs, we conducted a simulation-based experiment for a number of flash flood events in the West Africa region. QPFs from GFS and a regional WRF model are used to force a distributed hydrologic model that has been developed for regional scale flash flood predictions. The skill of hydrologic forecasts is compared against reference simulations based on satellite-based (IMERG) precipitation forcing for ~20 selected events and areas for which flash flood impacts have been reported in the region. Results demonstrate the limitations of coarse global QPFs relative to higher-resolution regional QPFs and provide important information for the development of operational flash flood forecasting systems in the region.

Xi Zhang

Senior Doctoral Student of Electrical Engineering

My research interests include point process analysis, modeling and optimization.

Akshay Aravamudan

Doctoral Student of Computer Engineering

Georgios C. Anagnostopoulos

Associate Professor of Electrical & Computer Engineering

I lead the Machine Learning Research Group at FIT.