Land Degradation Detection in Urban Areas Using Spatial Modelling and Semi-Automatic Classification of Satellite Imagery Data

2025-08-14T23:56:02+00:00

Urban land degradation poses a growing challenge in rapidly developing countries like Indonesia, where population growth and limited space drive uncontrolled land cover changes. This study aims to detect land degradation in urban areas through spatial modelling and semi-automatic classification of multi-temporal remote sensing imagery. Landsat-5 Thematic Mapper (TM) image from year 2011 and Landsat-9 Operational Land Imager collection 2 (OLI-2) image from year 2023 data were acquired from the The United States Geological Survey (USGS). Image pre-processing included band stacking, subsetting, and enhancement to improve visual interpretation. Semi-automatic supervised classification was applied to map seven land cover classes: agricultural dry land, rice field, forest, plantation, non-agricultural land, water body, and settlement. Training data and validation were supported by Google Earth Pro, official sources, and field surveys using random sampling. Change detection analysis revealed a 1664.65 ha increase in industrial areas, accompanied by significant reductions in rice fields (−1726.92 ha) and dry farmland (−1644.57 ha). The classification accuracy reached 80.24% and 75.11%, with kappa coefficients of 0.76 and 0.65, respectively. Results indicate that urban expansion is a key driver of land degradation, particularly through the loss of productive agricultural land. This research demonstrates the effectiveness of remote sensing-based spatial modelling and classification techniques for monitoring urban land degradation and informing sustainable land use planning.

Tropical Aquatic and Soil Pollution

Land Degradation Detection in Urban Areas Using Spatial Modelling and Semi-Automatic Classification of Satellite Imagery Data