Chergi AI

Chergi AI is a NASA-powered digital twin that turns Earth data into real-time city insights. Planners drag & drop parks, clinics, or cool roofs and instantly see heat, air, and equity impacts.

Comment

Chergi AI - NASA-Powered Healthy City Digital Twin

Inspiration

We were inspired by a simple idea: cities should be able to see the real impact of their decisions before they build anything. NASA collects world-class Earth data-heat, air quality, vegetation, night lights-but planners rarely get tools that turn this into clear, actionable insight.

Chergi AI was born from this gap. We wanted to turn raw satellite data into a visual, simulation-ready digital twin that helps improve human wellbeing and urban resilience.

What it does

Chergi AI creates an interactive 3D digital twin of a city using NASA data. It maps heat islands, air pollution, vegetation, infrastructure, and vulnerability.

Users can drag & drop interventions (new parks, clinics, cool roofs, traffic-reduction zones) onto the map. The system instantly updates metrics including:

  • Land Surface Temperature
  • NDVI and green access
  • NO₂ exposure
  • Healthcare accessibility
  • Impacted population
  • A real-time Healthy City Score

The Healthy City Score is calculated as:

$$ HCS = 100 - (w_h H + w_a A) + (w_g G + w_c C) $$

and the equity adjustment is:

$$ HCS_{eq} = HCS \cdot (1 - 0.4V) $$

The AI module then translates these changes into short, human-readable insights grounded in NASA science.

How we built it

  • Data: MODIS & Landsat LST, Sentinel-2 NDVI, TEMPO/OMI NO₂, VIIRS night lights, SEDAC/GHSL population and vulnerability
  • Backend: FastAPI for scoring, simulation, and raster tile generation
  • Math model: Standardization, weighting, and dynamic recomputation of metrics
  • Frontend: Next.js, Mapbox GL, and deck.gl for 3D visualization and drawing tools
  • AI module: Converts numerical deltas into natural-language explanations using structure-aware prompts and the equations above

Challenges we ran into

  • Cleaning and harmonizing multi-resolution NASA datasets
  • Building a fast raster tiling system that renders smoothly in the browser
  • Balancing scientific accuracy with real-time computational performance
  • Designing a scoring model that is both intuitive and rigorous
  • Ensuring AI explanations stay grounded in the actual math, not hallucination

Accomplishments that we're proud of

  • A fully working NASA-powered city digital twin
  • Real-time drag-and-drop planning simulations
  • A transparent Healthy City Score with equity built in
  • A smooth 3D UI that feels like a production planning tool
  • A clear link between NASA Earth observations and human wellbeing

What we learned

  • How to preprocess and tile NASA satellite data efficiently
  • How environmental indicators like heat, vegetation, and pollution interact in cities
  • How to design scoring systems that balance interpretability and rigor
  • How to integrate AI explanations without losing scientific grounding
  • How small interventions can meaningfully help vulnerable communities

What’s next for Chergi AI

  • Expand to more cities with automated global preprocessing
  • Add new interventions (solar rooftops, bus lines, EV corridors, reflective paving)
  • Integrate real-time weather, smoke, and flood layers
  • Release an open API for researchers and governments
  • Extend the system toward Moon/Mars settlement planning, adapting the digital twin for future space habitats
Share this project:

Updates