Monitoring and Forecasting Floods over North Africa based on Satellite data: Uncertainties and Challenges

North African hydrological processes undergo very high spatial and temporal variability due to tropical-subtropical interaction, land-sea contrast, complex orographic features, and gradients in soil moisture and temperature. The hydrological regimes of the region include major water basins in Tunisia and Morocco, which are generally characterized by low daily (discharge) flows that are disrupted by flash floods, so monitoring and predicting these events are complex unless we get reliable observations of rainfall, evapotranspiration and elevation. Multi-sensor satellite remote sensing along with rainfall-runoff routing models offer great potential to better understand the hydrological processes over North Africa. The overall objective of this presentation is to demonstrate how satellite data can be utilized to monitor and predict flood events over North Africa. As a first step, we have analyzed the major forcing hydrological data including ground measurements, satellite precipitation products (TMPA-RT, CMORPH and PERSIANN), and evapotranspiration data (USDA-ALEXI and MODIS). The uncertainties in the hydrological model estimates and their sensitivity to model inputs will be explained by selecting episodic floods over North Africa. This study also explores the multi-sensor data to better understand floods and flood frequency for different temporal aggregations by using the first principles of rainfall-runoff relationship (runoff fraction, lag time) and residual estimates of hydrological components based on water balance for the region. The research and techniques discussed here have application to many basins in arid regions, and can be particularly useful for transboundary basins where in-situ measurements are limited. Finally, we explain briefly the plausible dynamics involved in the occurrences of floods and the complexity in estimating the hydrological parameters from satellite data and models for North Africa.