Performance of GloFAS Flood Forecasts using proxy data in Uganda

Capabilities to forecast fluvial flooding are not equality spread across the globe and forecasting systems are especially limited in flood-prone low-income countries (Revilla Romero et al., 2014). The availability of higher spatial and temporal resolution remote sensing data and the increase in post processing technology have opened opportunities for fluvial forecasting at a continental and global scale (Emerton et al., 2016a) (Revilla-Romero et al., 2015). This means flood forecasts are available for regions where previously there were no forecasting capabilities. The availability of flood forecasts for flood-prone low-income countries does not directly lead to action being taken in case of flooding. The forecast based financing program of the Red Cross Climate Centre enables early action to be taken using probabilistic forecast information, with the aim of reducing the impacts of flooding (Coughlan de Perez et al 2015). The program uses a combination of forecast models including the Global Flood Awareness System (GLoFAS) and is active in multiple location including Tongo, Peru and Uganda. There are many factors at play to create an effective early warning system, including the performance of the forecast. Analysing the performance of forecasts is essential for the further improvement and development of an effective early warning system. However, in low-income countries with a low data availability this is a major challenge. This poster shows the performance of the GloFAS forecast using proxy flood event data in the North East of Uganda and poses the question: “How can the performance of forecasts be analysed when data is limited and uncertain?”