Spatio-temporal Analysis of Land Use/Cover Change (LUCC) Using Back Propagation Neural Network: A Case Study of Fuyang City, China

To address the limitations of low accuracy and error-prone classification in traditional maximum likelihood methods, this study uses three Landsat images from Fuyang City, China (2015, 2019, and 2024) as data sources. A Back Propagation (BP) neural network model is employed to conduct spatial analysis and quantitative research on high-precision LUCC derived from simulation predictions. First, the data were preprocessed, and training sample sets and test sets were selected with the assistance of ENVI software, followed by normalization of the dataset. Second, a BP neural network recognition model was constructed using MATLAB software, and the processes of model training, accuracy testing, and simulation prediction were sequentially performed. Subsequently, after applying plurality filtering and spatial raster analysis, the corrected classification data were imported into the ARCGIS platform for further spatial analysis. Finally, the study discusses the quantitative characteristics and ecological impacts of LUCC across the three phases in Fuyang City.