About the Project

AI-Driven Pollution Source Identification, Forecasting & Policy Dashboard

A comprehensive platform addressing Delhi-NCR's critical air pollution challenges through data science and AI.

Problem Statement

Delhi-NCR pollution is episodic and seasonal, driven by multiple overlapping factors: crop stubble burning in Punjab/Haryana, traffic congestion, industrial activity, and meteorological conditions. Current forecasting tools lack source-level granularity, citizen-level accessibility, and real-time policy feedback.

Our Solution

We developed an integrated software platform that provides:

  • Source Identification - Using satellite data (NASA MODIS, ISRO), CPCB monitors, and IoT sensors to trace pollution back to its source with high accuracy.
  • AI Forecasting - Machine learning-driven short-term (24-72 hours) and seasonal predictions for AQI trends.
  • Citizen App - Hyperlocal AQI with personalized health alerts, safe-route suggestions for commuting and outdoor activities.
  • Policy Dashboard - Data visualization tools for policymakers showing source contribution breakdowns, effectiveness of interventions, and AI-generated recommendations.

Expected Impact

Our platform aims to:

  • Provide transparent, data-driven insights for both citizens and policymakers
  • Help enforce evidence-based interventions (e.g., focus on stubble burning vs. vehicular restrictions)
  • Promote citizen engagement through personalized alerts and awareness
  • Contribute to long-term pollution reduction strategies in Delhi-NCR
Data Sources & Methodology
Primary Data Sources
  • NASA FIRMS - Satellite data for fire detection and monitoring
  • AQICN API - Real-time air quality index data
  • OpenWeather API - Meteorological data for forecasting models
  • CPCB - Government monitoring station data
  • ISRO - Indian satellite data for regional analysis
AI/ML Methodology

Our forecasting models use:

  • Time-series analysis (ARIMA, Prophet) for seasonal trends
  • Random Forest and Gradient Boosting for source attribution
  • Neural networks for complex pattern recognition
  • Ensemble methods to improve prediction accuracy
Development Team
Team Member
Rahul Sharma

Project Lead & Data Scientist

Specialized in environmental data analysis and machine learning applications.

Team Member
Priya Patel

Frontend Developer

Expert in responsive web design and data visualization techniques.

Team Member
Amit Kumar

Backend Developer

Focused on API integration and database management for real-time data processing.

Team Member
Neha Gupta

UI/UX Designer

Creating intuitive user experiences for both citizen and policymaker dashboards.

Technology Stack
Frontend
HTML5 CSS3 JavaScript Bootstrap 5 Chart.js
Backend
Python Flask REST APIs SQLite
Data Science
Pandas NumPy Scikit-learn TensorFlow
APIs Integrated
NASA FIRMS AQICN OpenWeather Google Maps
Smart India Hackathon 2025

This project is developed for the Smart India Hackathon 2025 under the theme Clean & Green Technology.

Problem Statement ID: 25216

Organization: AICTE

Department: AICTE

Category: Software