A new paper by JBA’s Murray Dale and Kay Shelton shows how probabilistic forecasting and local collaboration are supporting real-time flood alerts in Freetown, Sierra Leone. The prototype system combines high-resolution modelling with local data to strengthen preparedness in fast-responding basins.
Ready to discuss your next project? Contact our team today.
In Freetown, Sierra Leone, where intense rainfall can trigger dangerous floods within minutes, a new paper describes how advanced probabilistic weather forecasting tools can provide important flood risk guidance.
Authored by JBA’s Murray Dale and Kay Shelton, the paper illustrates how probabilistic forecasting, combined with local collaboration, has been used to develop actionable flood alerts in a capital city facing rapid-onset flood risk. It highlights the role of convection-permitting modelling (CPM) in managing hydrological extremes, using a prototype system developed for Sierra Leone as a case study.
By simulating convective rainfall more accurately, CPM supports real-time flood alerting through an ensemble prediction system (EPS). The paper highlights how this can then be used to develop messaging to strengthen preparedness in vulnerable, low-capacity settings.
Freetown, the country’s capital, faces frequent flash flooding due to intense rainfall, steep terrain, and less than ideal drainage infrastructure. These fast-responding basins leave little time for traditional flood warnings and pose serious risks to life and property.
From 2022 to 2024, JBA worked with Sierra Leonean agencies – including the National Meteorological Service, National Disaster Management Agency, and National Water Resources Management Agency – to co-develop a prototype flood alerting system, funded by the World Bank’s Global Facility for Disaster Reduction and Recovery.
The system makes use of the Weather Research and Forecasting (WRF) model, focused on the Freetown peninsula, integrating high-resolution rainfall predictions. It also used 15-years of local flood records and rain gauge data for calibration. Rainfall thresholds derived from this dataset enable the model to assess whether forecast rainfall is likely to trigger low- or high-impact flooding.
During the 2023 rainy season, the prototype was operated experimentally, issuing daily guidance messages containing advice co-developed with local agencies. These communicated the probability of flooding in clear terms – such as “medium chance of low-impact flooding” aligned to pre-agreed actions, helping agencies make proportionate responses.
“Developing this system with our partners in Sierra Leone has shown how probabilistic forecasting and context-specific solutions can make a real difference in fast-responding basins. By using suitable technology to co-develop an alerting and guidance system, we’re helping communities like Freetown better anticipate and respond to extreme weather events.”
Murray Dale, Technical Director, JBA.
As climate-related hydrological extremes become more frequent and severe, probabilistic forecasting offers a powerful tool for enhancing preparedness and resilience in vulnerable communities. The authors look forward to further advancing these approaches, supporting data-driven decision-making across sectors and geographies.
Read the full research article here.
Ready to discuss your next project? Contact our team today.
Discover more about the latest developments shaping resilience and driving sustainable development worldwide.
