Forecasting Flash Flood Impacts with IMERG-based Streamflow Simulations in Ecuador

Abstract:

Hydrologic models with the capability to utilize state-of-the-art satellite precipitation forcings have been limited in their application towards flash flood forecasting because of precipitation products latency or resolution. In this work, the Ensemble Framework for Flash Flood Forecasting (EF5) hydrologic framework, a modeling core used in the US National Weather Service for flash flood warning, is combined with GPM-IMERG data over Ecuador for the period 2000-2019 in order to explore its capability to capture and characterize the occurrence and severity of flash floods across the country. This information can prove valuable in developing new Forecast-based Financing (FbF) models for countries with larger numbers of ungauged basins, a core tactic utilized by the Red Cross in order to allocate resources in advance of natural disasters across the world. Reference stream data was provided by both the National Meteorological Agency of Ecuador (INAMHI) and the Global Data Runoff Centre. The evaluation approach uses both a traditional skill assessment of hydrologic performance (e.g NSCE, bias, RMSE) as well as a non-traditional assessment of model skill based off local event-based reporting data. The ability of the IMERG-forced modeling framework to capture useful information about events and their associated impacts is assessed, as well as the applicability of this same methodology towards regional, continental, and global flash flood forecasting systems.