Shalu

GE4050

PGE21320

This project aims to identify areas with the best opportunities for forest restoration using remote sensing and geospatial data. To identify potential restoration opportunities, the LandTrendr algorithm and satellite data analyze forest disturbance, recovery, and areas with no change. Multiple factors contribute to forest thinning, including climate, soil properties, and human intervention. A simple recurrent network model is developed to predict whether degraded forest pixels have restoration potential, achieving a classification accuracy of approximately 83%. The model is applied to the Ujjain district to identify regions with restoration potential. The study assesses the restoration potential of pixels that have experienced forest loss in the past 20 years.


Report Content

Forest degradation and global scenario

Forest restoration and activities

Problem statement & methodology of the project

Datasets, their relevance and LandTrendr parameters

Forest gain and loss change map

LandTrendr historical trend for the restoration sites

Data analysis and model architecture

Model justification and performance

Model application and related statistics

Conclusions and Limitations