Cool Durham: Using AI to Map Heat & Build Resilience
Inspiration
We were inspired to create Cool Durham because three out of four of us are data science majors who wanted to apply our skills to a local climate challenge.
Durham’s rising urban heat and its proximity to Chapel Hill, where we study, made it the perfect opportunity to build a project with real community impact.
What We Learned
We learned there is a gap between how people connect with data and the technologies or methods available to analyze it.
Even with rich datasets, turning data into meaningful insight can be extremely difficult.
This experience showed us how important it is to bridge that gap through visualization and AI tools.
How We Built It
We used data from the Data-Driven EnviroLab to build an AI-powered, interactive heat resilience dashboard for Durham.
- Loaded GeoTIFF raster temperature data into Python using Rasterio.
- Created a heat map visualization to show temperature variation across Durham.
- Tried to overlay the heat map on an interactive map using Folium and spent hours debugging when it would not display properly.
- Finally got it working by converting the raster data into a PNG image and providing it as input to Folium.
- Next steps include embedding the interactive heat map into a website and integrating a Gemini AI chatbot that explains patterns and suggests resilience strategies.
Challenges
- The initial dataset was too large to process efficiently.
- Folium would not overlay the heat map, leading to a 3-hour debugging session with researchers from the EnviroLab during office hours.
- The fix came after we realized we needed to convert the raster data into a PNG before overlaying it.
Takeaway
We learned the power of combining data science, AI, and local engagement to tackle climate issues.
By transforming complex temperature data into clear, actionable visuals, Cool Durham helps residents and policymakers build a cooler, fairer, and more resilient city.
(The demo video is listed under "Try it out" links since we couldn't upload to the main link.)

Log in or sign up for Devpost to join the conversation.