urbanalysis.earth

Integrating NASA and ESA Satellite Intelligence for Greener, Cooler, Cleaner Cities

This project was developed by Team Codenebula as part of the NASA Space Apps Challenge 2025.

It is an intelligent Smart Urban Planning Web Application that helps planners and policymakers analyze and mitigate three major urban environmental challenges:

• 🌳 Green Space Deficiency Analysis
• 🌡️ Urban Heat Island (UHI) Detection & Analysis
• 🌫️ Air Quality Hotspot Identification

Through the fusion of Earth Observation (EO) data from NASA and ESA satellites with AI-powered analytical frameworks, the platform enables data-driven, evidence-based, and scalable urban decision-making.

• Overview
• Core Functionalities
• How It Works
  • Green Access Analysis
  • Urban Heat Island Analysis
  • Air Quality Analysis
• NASA & ESA Data Sources
• Local Deployment Guide
• Acknowledgments
• License

Our system transforms complex satellite datasets into actionable, location-specific insights using a combination of geospatial analytics and AI-powered natural language generation:

1. Urban Heat Island (UHI) Analysis – Detect and map high-temperature clusters using MODIS and ECOSTRESS LST data.
2. Air Quality Analysis – Identify NO₂ and PM₂.₅ concentration hotspots using Sentinel-5P and GEOS-CF data.
3. Green Access Analysis – Quantify vegetation distribution and detect “green space gaps” using Sentinel-2 NDVI and OpenStreetMap park data.
4. AI Decision Support – The system uses a Large Language Model (LLM) to translate technical analyses into human-readable planning advice.
5. Interactive Visualization – Web-based, map-centric interface for planners to explore layers, hotspots, and AI recommendations interactively.

Objective: Detect vegetation-deficient neighborhoods and recommend new micro-parks. Pipeline:

• Retrieve Sentinel-2 surface reflectance (Bands 8 & 4).
• Compute NDVI = (NIR – Red) / (NIR + Red).
• Classify NDVI into low (< 0.2), moderate (0.2–0.5), and high (> 0.5) greenness zones.
• Overlay with OpenStreetMap (OSM) data for roads, parks, and accessibility.
• Identify underserved zones beyond a 10-minute walking radius from existing parks using OSMnx isochrones.
• Feed contextual data (population, soil, terrain) into the AI engine for micro-park recommendations.

Output: Interactive NDVI maps, green gap overlays, and AI-generated “Suitability & Recommendation” briefs .

Objective: Detect heat stress zones for cooling interventions. Pipeline:

• Acquire MODIS day/night and ECOSTRESS Land Surface Temperature (LST) data.
• Compute z-score anomalies relative to city means.
• Apply DBSCAN clustering to identify contiguous heat clusters (> 2σ).
• Integrate land cover, NDVI, and population exposure data.

Output: Severity-coded UHI maps (Elevated / High / Severe) with tables showing population exposure and recommended cooling measures (e.g., green roofs, reflective surfaces, tree corridors) .

Objective: Locate air pollution hotspots and vulnerable communities. Pipeline:

• Aggregate Sentinel-5P NO₂ and GEOS-CF PM₂.₅ reanalysis data.
• Standardize via z-scores and apply DBSCAN clustering to delineate hotspots.
• Enrich with contextual layers (population, schools, hospitals, industrial zones).

Output: Pollution cluster maps, exposure summaries, and AI briefs linking pollution with land use and policy interventions .

• RAM: ≥ 4 GB

• Processor: Any modern multi-core CPU

• Python Version: 3.11.9

• APIs:

  • Google Earth Engine API
  • Groq API

The system runs with two backend servers. One liteweight server for the UI interface controll and another for running the analysis. The UI server communicates with the analysis server.

Client <---> UI Server <---> Analysis server

1. Clone the repository

   git clone https://github.com/IamDipankar/urbanalysis.git
   cd urbanalysis

2. Create a Conda virtual environment

   conda create -n urbanenv python=3.11.9

3. Activate the environment

   conda activate urbanenv

4. Install dependencies

   pip install -r requirements.txt

5. Create a .env file

   Create a .env file in the root directory with the following configuration:

   EE_SERVICE_ACCOUNT=<Your earth engine service account>
   EE_KEY_B64=<Your earth engine key json file base64 encrypted>
   GROQ_API_KEY=<Your groq api key>
   REMOTE_SERVER_URL=<Remote server / Analysis server URL>
   MAIN_SERVER_URL=<Main server / UI server URL>

6. Run the application

   uvicorn main:app --host 0.0.0.0 --workers 1 --port <Give a port name>

   Open a different terminal instance and run:

   uvicorn remotemain:app --host 0.0.0.0 --workers 1 --port <Different port name>

   Here an url will be generated for each server depending on port and host. This urls should match with the .env file.

7. Load the web app Click on the link shown in your console (eg. http://127.0.0.1:8000 or localhost:8000) to open this site on your browser.

Team Codenebula

We gratefully acknowledge:

• NASA and ESA for providing open-access satellite datasets (MODIS, ECOSTRESS, Sentinel-2, Sentinel-5P).
• Google Earth Engine for geospatial data access and processing.
• OpenStreetMap for open geographic data.
• FastAPI for backend development.
• OSMnx for network and accessibility analysis.
• The AI and data science community for continuous innovation in open geospatial intelligence.