GreenCity.ai

Inspiration

Cities are on the front lines of climate change, but the data needed to plan effective green infrastructure is often inaccessible, fragmented, or difficult to interpret. Urban heat disproportionately affects vulnerable communities, yet many planners lack easy tools to identify where interventions like trees, green roofs, or rain gardens would have the greatest impact. We were inspired to bridge this gap by turning satellite data into something actionable, intuitive, and city-specific—so climate resilience planning can be informed by evidence rather than guesswork.

What it does

GreenCity.ai is an interactive urban climate analysis platform that identifies heat-vulnerable areas within cities. Users enter a city name, and the system automatically determines the city’s geographic extent, retrieves NASA satellite land surface temperature data, and visualizes urban heat patterns on an interactive map. The platform highlights heat hotspots and allows users to explore where green infrastructure investments could most effectively reduce heat stress. The system is designed to support future layers such as flood risk and air pollution to enable holistic green infrastructure planning.

How we built it

We built GreenCity.ai using a combination of AI, geospatial processing, and satellite remote sensing. An AI model is used to infer a city’s bounding box from a simple city name input. This bounding box is then used to query NASA Earthdata and retrieve MODIS Land Surface Temperature granules. The data is processed by mosaicking multiple satellite tiles, clipping them to the city boundary, and converting them into analysis-ready geospatial formats. We built an interactive front-end using Streamlit and Folium to render heat maps, identify hotspots, and provide a user-friendly interface for exploration.

Challenges we ran into

One of the biggest challenges was working with real satellite data pipelines under time constraints. Satellite products are tiled, large, and often stored in specialized formats, requiring careful handling of coordinate reference systems, mosaicking, and clipping. We also encountered API rate limits and service availability issues when integrating AI and data download services, which required caching, retries, and graceful fallbacks. Designing an automated pipeline that remained stable under Streamlit’s reactive execution model was another key technical challenge.

Accomplishments that we're proud of

We’re proud of building a fully automated, end-to-end pipeline that goes from a city name to a scientifically grounded heat map using real NASA satellite data. The platform integrates AI-driven geospatial reasoning, remote sensing data retrieval, and interactive visualization in a way that is both technically robust and easy to use. Most importantly, the system is extensible and designed with real-world urban planning applications in mind, not just a static demo.

What we learned

Through this project, we gained hands-on experience working with Earth observation data, geospatial transformations, and large-scale data pipelines. We learned how to design resilient systems that handle unreliable external services and how to manage state and caching in reactive web applications. The project also reinforced the importance of clear abstractions when combining AI, scientific data, and user interfaces into a single coherent system.

What's next for GreenCity.ai

Next, we plan to expand GreenCity.ai beyond heat analysis by incorporating flood risk, air quality, and land-cover data to provide multi-criteria recommendations for green infrastructure placement. We also aim to add neighborhood-level summaries, equity-focused metrics, and policy-friendly reports that planners can directly use in decision-making. Ultimately, we envision GreenCity.ai as a scalable decision-support tool that helps cities design more resilient, equitable, and climate-adaptive urban environments.

Built With

• python
• streamlit