Suraj Gharde

All projects

JalNetra जलनेत्र

Satellite-Based Water Quality Monitoring Platform

Code on GitHub(opens in new tab)

Key numbers

  • 13 GB to 7 MB data per satellite pass
  • 30 water bodies in Maharashtra
  • 75 alerts from 2.5 years of data

How it works

  1. 1Input

    Sentinel-2 imagery for 30 water bodies in Maharashtra

  2. 2Process

    Celery and Earth Engine pipeline: 13 GB to 7 MB per satellite pass

  3. 3Detect

    ML anomaly detection (DBSCAN, Isolation Forest, seasonal baselines) with explainable scoring

  4. 4Output

    75 alerts from 2.5 years of data, on a map dashboard and a GPT-powered Telegram bot

What I built

  • Reduced data per satellite pass from 13 GB to 7 MB with a Celery and Earth Engine pipeline analyzing Sentinel-2 imagery for 30 water bodies in Maharashtra.
  • Generated 75 alerts from 2.5 years of data using ML anomaly detection (DBSCAN, Isolation Forest, seasonal baselines) and explainable scoring, surfaced through a map dashboard and a GPT-powered Telegram bot.

Stack

Python, FastAPI, React.js, PostgreSQL, Google Earth Engine

Celery, DBSCAN, Isolation Forest