Document Type : Original Article

Authors

• Sadra Karimzadeh ¹
• Abed Touma Jasim ²
• Abolfazl Ghanbari ³
• Muntadher Aidi Shareef ⁴

¹ Department of Remote Sensing and GIS, University of Tabriz, Iran

² , Department of Remote Sensing and GIS, Faculty of Planning and Environmental Sciences, University of Tabriz, Tabriz, Iran

³ Department of Remote Sensing and GIS, Faculty of Planning and Environmental Sciences, University of Tabriz, Tabriz, Iran

⁴ Department of Surveying Technical Engineering, Faculty Technical Engineering, Northern Technical University, Kirkuk 36001, Iraq

Abstract

Monitoring land-use/land-cover (LULC) dynamics is essential for understanding environmental change and supporting sustainable land management in semi-arid regions. This study evaluates feature-level Optical–SAR fusion by integrating Landsat multispectral imagery with Sentinel-1 VV and VH backscatter data for LULC classification and change detection in Kirkuk City, Iraq, between 2015 and 2024. Five land-cover classes—desert, barren land, agriculture, water, and built-up areas—were classified using a Support Vector Machine (SVM) classifier and assessed with overall accuracy (OA), Kappa coefficient, producer’s and user’s accuracies, and confusion matrices. The Landsat + Sentinel-1 VV fusion dataset achieved the highest classification accuracy in 2015 (OA = 96.51%, Kappa = 0.954). However, by 2024, increasing landscape complexity reduced classification performance for both fusion datasets. Change detection revealed substantial environmental transformations, with agricultural degradation as the dominant transition. Agricultural land decreased by 45.40% and 30.15%, primarily converting to barren land in the VH and VV fusion datasets, respectively. Additional transitions from agriculture and barren land to desert indicate accelerating desertification, while built-up areas showed persistent expansion, largely at the expense of agricultural land. Confusion matrix analysis also revealed increasing spectral similarity between barren land and desert, suggesting the emergence of transitional degradation zones. Overall, the findings identify agricultural degradation, desertification, and urban expansion as the principal environmental processes shaping the study area and demonstrate the value of Sentinel-1 VV and VH data integrated with Landsat imagery for operational LULC monitoring in semi-arid environments.

Keywords

• Sentinel 1 SAR
• Image Fusion
• LULC
• Desertification
• agricultural degradation
• urban expansion

Main Subjects

• Remote sensing and Gis