The NFLICS Project: Nowcasting Flood Impacts of Convective Storms in the Sahel

Abstract:

Flash flooding from intense rainfall frequently results in major damage and loss of life across Africa. In the Sahel, intense rainfall from Mesoscale Convective Systems (MCSs) is the main driver of flash floods, with recent research showing that these have tripled in frequency over the last 35 years. This climate-change signal, and rapid urban expansion in the region, suggest that the socio-economic impacts of flash flooding will become more frequent and severe. Appropriate disaster preparedness, response, and resilience measures are required to manage this increasing risk.

The project NFLICS (Nowcasting FLood Impacts of Convective storms in the Sahel) is developing a prototype early warning system for Senegal, nowcasting flood risk from MCSs at city and sub-national scales. To identify convective activity, wavelet analysis is applied to Meteosat data on cloud-top temperature from 2004 to 2019, and for the start-time of a nowcast. Data on historical convective activity, conditioned on the present location and timing of observed convection, are used to produce probabilistic forecasts of convective activity out to 6 hours ahead. Verification against the forecast analysis and 24-hour raingauge accumulations over Dakar suggests that these probabilistic nowcasts provide useful information on the occurrence of convective activity. The highest skill (when compared to forecasts based on climatology) is obtained when the probability of convection is estimated over spatial scales of 100 to 200km.

A flood knowledge database, compiled with local partners, allows estimation of the flood risk over Dakar given the identified probability of convective activity. Flood hazard is estimated from the probabilistic convective-activity nowcast, combined with information on the historical relationship between convective activity and precipitation totals. Information on antecedent conditions is also included, with a higher level of hazard associated with recent rainfall and already-wet conditions. Flood vulnerability is estimated at the local scale from post-event analysis of the 2009 floods along with information from recent modelling studies and flood-alleviation measures. The combined information from nowcasts of convective-activity and flood risk is visualised through an interactive GUI.