Tree cover loss: Madhya Pradesh becomes first state to launch AI-driven forest alert system

Representative image

Madhya Pradesh has become the first state in the country to implement an Artificial Intelligence (AI)-based real-time forest alert system designed for active forest management.

The initiative, which integrates satellite imagery, machine learning, and mobile-based field feedback, enables prompt detection of illegal land encroachments, unauthorised land use changes, and forest degradation.

The AI-driven system, launched as a pilot project, is currently operational in five sensitive forest divisions, which includes Shivpuri, Guna, Vidisha, Burhanpur, and Khandwa. Notably, all these five regions are known for frequent incidents of illegal encroachment and tree felling.

The real-time forest alert system utilises Google Earth Engine to analyse multi-temporal satellite imagery and applies a custom-built AI model to detect anomalies in land use. It triggers an alert when a change is detected, such as new construction, tree cover loss, or cropping activity.

The alert is instantly dispatched to field staff through a dedicated mobile app. Staff are then expected to verify the alert on-site and provide feedback, including GPS-tagged photographs, voice notes, and comments, enabling real-time monitoring and rapid response.

Guna Divisional Forest Officer (DFO) Akshay Rathore, who conceptualised the AI system for forest management, said this is the first time the forest department has combined satellite, AI, and field feedback in a continuous cycle, which improves itself over time.

Each alert contains more than 20 data attributes, including polygon-based change detection, Normalised Difference Vegetation Index (NDVI), Soil Adjusted Vegetation Index (SAVI), Enhanced Vegetation Index (EVI), and SAR attributes. These metrics help refine the precision of alerts and inform better decision-making.

"This system empowers forest staff to monitor and take immediate action. Alert generation and feedback process includes initial alert generation using Google Earth Engine to compare satellite images of three dates, identify changes in crops, wasteland, construction etc," Rathore said.