article · Ecology and Evolution
An evaluation of the Billion Tree Afforestation Project in Pakistan's Khyber Pakhtunkhwa province demonstrates substantial ecological recovery between 2015 and 2023. By combining Sentinel-2 satellite imagery with Random Forest machine learning classification, tree cover was shown to expand from 25.02 percent to 29.99 percent, while barren land decreased from 20.64 percent to 16.81 percent, achieving an overall classification accuracy exceeding 85 percent. Spatial clustering and hotspot analyses indicated an increase in high-confidence vegetation recovery zones from 36.76 percent to 42.56 percent. Furthermore, an artificial neural network model accurately predicted changes in the Normalized Difference Vegetation Index with high statistical reliability, while feature attribution identified precipitation and soil moisture as the primary drivers of plant growth. These methods provide a reliable analytical framework to monitor large-scale planting initiatives and guide long-term environmental management programmes.
Large-scale tree planting programmes require rigorous verification to prove their ecological impact. By combining satellite data with machine learning, environmental planners and policymakers can objectively track vegetation changes over time and understand the climate factors that sustain them. This provides an evidence-based approach to verify reforestation outcomes, optimise future planting strategies, and protect land resources more effectively.
The combination of satellite imagery, Random Forest classification, and neural network modelling offers an applied, tested framework for ecological monitoring. Environmental agencies, conservation bodies, and carbon-offset verification services could use these methods to independently audit large-scale land restoration projects. Because the approach relies on accessible Sentinel-2 imagery and established predictive modelling, it appears close to real-world operational deployment for forestry management and environmental compliance monitoring.
AI-generated from the published abstract. Always read the original work before citing.
This study evaluates the Billion Tree Afforestation Project (BTAP) in Pakistan's Khyber Pakhtunkhwa (KPK) province using remote sensing and machine learning. Applying Random Forest (RF) classification to Sentinel-2 imagery, we observed an increase in tree cover from 25.02% in 2015 to 29.99% in 2023 and a decrease in barren land from 20.64% to 16.81%, with an accuracy above 85%. Hotspot and spatial clustering analyses revealed significant vegetation recovery, with high-confidence hotspots rising from 36.76% to 42.56%. A predictive model for the Normalized Difference Vegetation Index (NDVI), supported by SHAP analysis, identified soil moisture and precipitation as primary drivers of vegetation growth, with the ANN model achieving an <i>R</i> <sup>2</sup> of 0.8556 and an RMSE of 0.0607 on the testing dataset. These results demonstrate the effectiveness of integrating machine learning with remote sensing as a framework to support data-driven afforestation efforts and inform sustainable environmental management practices.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1002/ece3.70736
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
