How can Africa's flood risk be mapped with AI?
The Future Resilience for African Cities and Lands (FRACTAL) project’s Participatory Climate Information Distillation for Urban Flood Resilience in Lusaka (FRACTAL-PLUS) exemplifies climate services in action. By merging local knowledge with advanced analyses and technology, FRACTAL-PLUS enhances resilience strategies with contextual accuracy. Localising global rainfall data and GIS datasets with regional climate data to create relevant flood hazard maps, discussed and refined by end users, brings narrative fidelity to scalable climate services beyond academic and geographic boundaries.
Key to this was the learning lab format, initiated with evidence gathering and dialogue in late 2021. Twenty in-person attendees, including city stakeholders (e.g., Lusaka Water Security Initiative – LuWSI, Lusaka City Council), a community project team, and a UK Met Office representative, participated, with additional team members engaging virtually. The labs used interactive exercises, games, presentations, and discussions to reveal complexities in addressing urban flood risks under a changing climate. This collaborative effort united the lab participants to co-develop strategies for future adaptation and resilience. Two bridging surveys completed in February 2022, with 20 in-depth responses from flood-affected Lusaka residents, guided the strategic representation of flood risk across Lusaka’s diverse socio-economic settings.
This participatory approach generated valuable data, engaging local stakeholders meaningfully and ensuring relevant findings for future needs. The learning labs were also instrumental in gathering historical observations of flooding and climate change experiences for Lusaka, and in helping to distil community-based narratives of flood risk, resilience, and socio-economic vulnerability. These narratives added a human dimension to the technical data presented in the labs, making it more accessible and actionable. Using Natural Language Processing (NLP) and Text Network Analysis (TNA) through the open-source InfraNodus model visualised the large volume of text data generated by the learning labs, uncovering patterns, gaps, and insights from within the learning lab transcripts, providing a rapid processing format for experiential and sensory data into actionable goals around the themes of climate services.
Policymakers can use narrative-informed flood maps to identify high-risk areas and develop targeted interventions that align with local experiences. This approach has potential to enhance social resilience to floods and climate impacts across Africa. For climate services practitioners, this model fosters community engagement in data gathering and decision-making, ensuring scientifically robust and socially relevant outputs. The interactive learning labs fostered trust and collaboration, improving communication of climate risks and resilience strategies. Beyond flood resilience, these methodologies can address other climate challenges like droughts, heatwaves, and sea-level rise in different locations. Integrating local knowledge with advanced data analysis enhances resilience across various impacts with minimal resources. The FRACTAL-PLUS project therefore demonstrates and underscores the importance of interdisciplinary collaboration, uniting natural and social scientists to address the complexities of climate resilience, serving as a benchmark for future climate services research and practice.
This paper will demonstrate the power of combining local knowledge with advanced modelling tools to improve climate services through enhancing flood resilience. Localised flood maps informed by community narratives can act to enhance stakeholder engagement and address social vulnerability to climate change. The project’s success highlights the value of interdisciplinary collaboration and offers practical insights for policymakers and practitioners to enhance resilience and support communities facing climate change.
Publication: Q4/2025