Monitoring Forest Cover Dynamics Using Orthophotos and Satellite Imagery

The assessment of changes in forest coverage is crucial for managing protected areas, particularly the face climate change. This study monitored cover dynamics a 6535 ha mountain area located north-west Romania as part Apuseni Natural Park from 2003 to 2019. Two approaches were used: vectorization orthophotos and Google Earth images (in 2003, 2005, 2009, 2012, 2014, 2016, 2017, 2019) satellite imagery (Landsat 5 TM, 7 ETM, 8 OLI) pre-processed Surface Reflectance (SR) format same years. We employed four standard classifiers: Support Vector Machine (SVM), Random Forest (RF), Maximum Likelihood Classification (MLC), Spectral Angle Mapper (SAM), three combined methods: Linear Unmixing (LSU) with Breaks (NB), Otsu Method (OM) SVM, extract classify areas. Our had two objectives: 1) accurately assess over 17-year period 2) determine most efficient methods extracting classifying validated results using performance metrics that quantify both thematic spatial accuracy. indicate 9% loss area, representing 577 an average decrease ratio 33.9 ha/year−1. Of all used, SVM produced best (with score 88% Overall Quality (OQ)), followed by RF mean value 86% OQ).