Urban Heat Decision Support System

ISRO Bharatiya Antariksh Hackathon 2026

An AI-powered Urban Heat Decision Support System designed to identify urban heat hotspots, analyze their causes, simulate cooling strategies, and provide actionable recommendations for urban planners.

Project Vision

Urban areas experience increasing temperatures due to rapid urbanization, reduced vegetation, and heat-retaining infrastructure. This project combines satellite remote sensing, geospatial analysis, machine learning, and interactive visualization to build a decision-support platform for urban heat mitigation. The system aims to:

• Identify urban heat stress hotspots
• Predict Land Surface Temperature (LST)
• Explain major heat drivers
• Simulate cooling interventions
• Recommend optimized mitigation strategies

System Architecture

Satellite & Geospatial Data
        │
        ▼
Data Collection & Processing
        │
        ▼
Feature Engineering
NDVI
NDBI
LST
Vegetation
Elevation
Weather
Albedo
        │
        ▼
Machine Learning Model
(Random Forest / XGBoost)
        │
        ▼
Heat Prediction
        │
        ▼
Explainable AI
(SHAP + Feature Importance)
        │
        ▼
Scenario Simulation
(Tree Cover
Cool Roofs
Water Bodies)
        │
        ▼
Recommendation Engine
        │
        ▼
Interactive Decision Dashboard

Features

Interactive Heat Map

• Visualize satellite-derived environmental layers
• Explore Land Surface Temperature
• Compare multiple geospatial features

Feature Analysis

• Distribution analysis of environmental variables
• Feature statistics
• Correlation analysis
• Explanation of feature importance

AI Prediction

(Integration in progress)

Planned:

• Machine learning based heat prediction
• Predicted temperature maps
• Model performance metrics

Heat Driver Analysis

(Integration in progress)

Planned:

• SHAP explainability
• Feature importance ranking
• Understanding why an area is hot

Cooling Scenario Simulation

(UI implemented)

Planned interventions:

• Increase tree cover
• Cool roofs
• Reflective pavements
• Water body expansion

Recommendations

(UI implemented)

Provides future:

• Priority zones
• Cooling strategies
• Estimated impact

---

Dashboard Structure

The application is built using Streamlit.

Urban Heat Decision Support System
│
├──  Map
│
├──  Analysis
│
├──  Prediction
│
├──  Drivers
│
├──  Simulation
│
└──  Recommendations

---

Tech Stack

Frontend / Dashboard

• Streamlit
• Folium
• Plotly
• Matplotlib

Geospatial Processing

• Rasterio
• NumPy
• GeoTIFF data processing

Data Science

• Pandas
• Scikit-learn
• XGBoost (planned)

Explainable AI

• SHAP (planned)

---

Project Structure

isro/
│
├── app.py
│
├── components/
│ ├── analysis.py
│ ├── map.py
│ ├── prediction.py
│ ├── drivers.py
│ ├── simulation.py
│ └── recommendations.py
│
├── utils/
│ ├── raster.py
│ └── constants.py
│
├── data/
│ └── (local datasets)
│
├── assets/
│
└── README.md

---

Installation

Clone the repository:

git clone <repository-url>

Create virtual environment:

python -m venv .venv

Activate environment: Mac/Linux:

source .venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Run application:

streamlit run app.py

Current Progress

• ✅ Streamlit application structure
• ✅ Tab-based dashboard layout
• ✅ Real geospatial feature loading
• ✅ LST visualization
• ✅ Feature analysis module
• ✅ Simulation interface
• ✅ Recommendation interface
• 🚧 Machine learning integration
• 🚧 SHAP explainability
• 🚧 Cooling simulation engine
• 🚧 Final map enhancement

Team Roles

AI & Integration

• Machine learning pipeline
• Model development
• System integration

Geospatial Data Engineering

• Satellite data processing
• Feature generation
• GIS layers

Urban Climate Analytics

• Heat driver analysis
• Explainable AI
• Intervention strategies

Dashboard & Visualization

• Streamlit interface
• Interactive visualization
• User experience

Future Enhancements

• Real-time satellite data updates
• Higher-resolution heat maps
• Automated intervention optimization
• City-scale deployment
• Web-based GIS interface

Hackathon

Developed for: ISRO Bharatiya Antariksh Hackathon 2026 Topic: Optimizing Urban Heat Mitigation and Cooling Strategies using AI/ML

THIS WAS MY CONTRIBUTION TO MY TEAM FOR HACKATHON WHEREIN I DEVELOPED THE FRONTEND DEMO

ANYUTA KUMAR