Akademik Veri Yönetim Sistemi
International Journal of Engineering and Geosciences, cilt.11, sa.1, ss.78-88, 2026 (ESCI, Scopus, TRDizin)
The study investigates changes in the İğneada Longoz forests in Kırklareli Province, Türkiye, between 2000 and 2021 using Landsat-7 and Landsat-8 satellite imagery. Classification was performed with the random forest (RF) machine learning algorithm and remote sensing (RS) techniques on the Google Earth Engine (GEE) platform. To better distinguish longoz forests and other forested areas from surrounding features, the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) were integrated into the classification alongside standard spectral bands. In the first stage of the classification process, the study area was categorized into seven classes, identifying all forest areas. In the second stage, the area was further divided into nine classes, isolating longoz forests from other forest types. Additionally, NDVI and EVI time series analyses were conducted to evaluate forest phenology. According to the classification results, the initial stage achieved overall accuracy ranging from 79.97% to 90.63%, with an average of approximately 85.00%. The Kappa statistics varied between 0.721 and 0.877. In the second stage, the overall accuracy ranged from 75.92% to 90.04%, while the Kappa statistic was between 0.649 and 0.866.