Approaches to community–scale drought and flood early warning

The African Flood and Drought Monitor (AFDM) is a satellite-model based system for monitoring and forecasting flood and drought conditions for the African continent. It has been running operationally since 2008 in various forms, providing useful information on flood and drought risks to a variety of end-users, as well as information for a wide range of water-related applications, including food and energy security, health risks, and migration. This paper provides an overview of the latest version of the system, which incorporates updated versions of hydrological models and meteorological data at high resolution, as well as the state-of-the-art short-term and seasonal forecast models. We also describe the tailoring of the AFDM to national systems across southern Africa and the process of co-design of these systems with national agencies and end-users. The system has been evaluated on several time scales, historically and for short-term (flood) and seasonal (drought) forecasts. Predictability is discussed with respect to end-users needs, especially at the community scale, and how more recent approaches to predict at the community scale are being incorporated into these monitoring systems, using high-performance computing, machine learning and data assimilation.