Published on 23 September 2025
By late morning, heat builds up in the streets of Nairobi’s informal settlements, warming the walls and lingering at breathing height. Climate change is often described in global averages, but its effects are lived locally - varying street by street, house by house. In informal settlements with metal roofs and narrow lanes, heat strikes hard: metal sheets absorb the sun’s energy, bounce sunlight onto neighbouring surfaces, and radiate stored heat back into the air. Trapped between close-set walls, these combined effects turn alleys into radiant corridors that amplify thermal stress for residents. Yet the thermometers that guide policy rarely capture these realities. Weather stations are typically located on airport lawns or university campuses, far from the metal roofs and smoky streets in densely built-up neighbourhoods. Satellites’ thermal sensors help analyse thermal inequalities but measure surface, not air, temperature – the temperature a child feels walking to school. If we don’t measure the heat people actually experience, we underestimate risk.
Measure and model air temperature in informal settlements
In ONEKANA, we developed a scalable method that cities everywhere can adopt to measure and model air temperature in informal settlements using community-based surveys and satellite data. working with local leaders, we designed walking routes that passed through critical community spaces such as water points, schools, markets, clinics, and sun-exposed street crossings. Residents then carried out two-hour campaigns during the hottest part of the day, using low-cost temperature sensors mounted on wooden poles to record air temperature along these routes. These ground measurements were combined with predictors derived from satellite imagery (e.g., ECOSTRESS LST, albedo, Sentinel-2 spectral indices) and building footprint metrics, in a machine-learning model to map the air temperature across the settlements.
Figure 1: A young girl walks to school in an informal settlement in Nairobi.
Where heat does the most harm and who is least able to cope?
Mapping air temperature alone doesn’t tell the whole story. Understanding heat risk also requires knowing how susceptibility varies within settlements: where heat does the most harm and who is least able to cope. To capture this, we co-designed household micro-surveys with community leaders to record housing materials (roof, ceiling, walls), reported heat difficulties (discomfort, sleep loss), household crowding, protective assets (natural ventilation, fans), and the access to basic services (potable water, electricity, waste management, healthcare). Residents collected responses using a mobile app. We then trained machine-learning models to predict a susceptibility index across all locations, extending estimates beyond surveyed sites using Sentinel-2 satellite data and other geospatial predictors.
Figure 2: Temperature survey conducted in Nairobi
Population distribution mapping
To quantify where residents are both highly exposed and most susceptible to heat, we mapped population distribution across the settlements using a deep-learning model trained on PlanetScope satellite imagery and community-survey inputs. We then combined these population maps with air-temperature and thermal-susceptibility layers to locate the most at-risk areas.
Figure 3: Population density map in the settlements.
Share the results with local people
Mitigation measures emerged from a stakeholder workshop with community representatives, NGOs, planners, municipal officials and academics. Project results were shared to spark debate on feasible, low-cost, and community-driven adaptations. Participants proposed targeted actions such as: planting trees and installing canopies to reduce afternoon heat; painting and insulating the hottest, most reflective metal roofs; improving ventilation in homes and community halls with ridge vents and cross-breezes while ensuring security; shading water points, clinic queues, and play areas; and establishing small cooling hubs with backup power for vulnerable groups in areas where heat persists.
Figure 4: Stakeholder workshop
In conclusion...
Modelling local heat variations with remote sensing and community data changes the conversation. The problem is no longer abstract but visible as hotspots that demand action. By generating and sharing evidence, we aim to help guide the next round of interventions. Seen at neighbourhood scale, climate action becomes concrete: making homes habitable, cooling streets, and planning with the worst-affected communities at the centre.
Project team
ULB: Eléonore Wolff - Sabine Vanhuysse - Stefanos Georganos - Angela Abascal
ITC: Monika Kuffer - Jon Wang
University of Navarra: Sally Sampson