Akademik Veri Yönetim Sistemi
European Journal of Forest Engineering, cilt.12, sa.1, ss.79-89, 2026 (Scopus)
Long-term land use and land cover (LULC) data are crucial for hydrological and forestry analyses, but CORINE Land Cover data are only available from 1990 onwards. This limitation prevents the assessment of earlier land cover changes, particularly in environmentally sensitive areas. To address this issue, we reconstructed historical LULC maps for the Eastern Black Sea Region of Türkiye for the years 1975, 1980, and 1985. Historical land cover conditions were reconstructed using Landsat MSS and TM images processed in the Google Earth Engine environment. To ensure consistency between sensors, summer season NDVI composites were produced and harmonized using a percentage-based normalization approach. Subsequently, land cover classification was performed using a Random Forest algorithm incorporating topographic variables and temporal pre-information obtained from post-1990 CORINE datasets, and pseudo-land cover maps compatible with CORINE were created. The reconstructed maps were evaluated through class distribution analysis, transition matrices, and stability measures. The findings reveal a high degree of consistency between the retrospectively estimated datasets and the official CORINE inventories, particularly in terms of dominant land cover layers. Furthermore, the transition dynamics and stability patterns observed in the reconstructed periods closely mirror those of the official time series, confirming that the methodology avoids creating misleading temporal artifacts. The proposed framework provides a reliable extension of CORINE-compliant land cover information to pre-1990 periods, establishing a robust foundation for long-term environmental, hydrological, and forestry studies.