Forecast production

Diagram about production sketch
Overview of the essential components and data flow of the operational hydrological forecasting and alert ICT system used in FANFAR. Circles represent data, and boxes represent processors (computational models, scripts, and human interpreters).
In FANFAR, hydrological forecasts and flood risk information are produced using an operational hydrological forecasting and alert pilot ICT system. The core of the system is a hydrological model, whose main function is to predict the effects of meteorological dynamics (e.g. rainfall and temperature) on river flow, water level, soil moisture in rivers, lakes, wetlands, and all land surface areas. In FANFAR, we use the Niger-HYPE model for the Niger River basin and the World-Wide HYPE model for the entire West African domain ( Two simulations are carried out to make a forecast with these hydrological models:
  1. In blue above: A simulation of a historic spin-up period up until the day before the forecast (t<0 → t0, t=time), producing a model state that represents the present hydrological conditions at the start of the forecast period (the initial state, t0). The core input for this simulation is meteorological ‘analysis’ data (i.e. fusion of meteorological observations and models representing historic conditions). Additional observations of the hydrological conditions (e.g. in-situ river flow observations or remotely sensed water levels) are gradually introduced in the system in order to improve the initial state through data assimilation.
  2. In orange above: A simulation for the forecast period relying on the initial state and meteorological forecast data about future weather dynamics (t1 → t>1). The output of the forecast simulation is a representation of future hydrological conditions.
The next step in the process (in green above) is to derive useful forecast and flood risk information. What is considered ‘useful’ clearly depends on the user, which is defined together during the FANFAR co-design workshops. Typically, a set of flood risk thresholds are specified based on historic conditions or local knowledge. The forecasts are then compared against these thresholds to determine the severity of the situation and the potential flood risk level. This information is finally delivered to users through various distribution channels (e.g. online visualisation or email and SMS to flood risk information subscribers). Operational forecasting heavily relies on scheduled, automated execution and monitoring of the data flows and processing tasks on an ICT environment that is always up and running (in grey above). To this end, FANFAR uses the Hydrology Thematic Exploitation Platform (Hydrology-TEP). Hydrology-TEP is a cloud-based ICT environment.