Change detection in satellite imagery plays a crucial role in environmental monitoring and spatial analysis. However, conventional methods often face limitations in processing efficiency and accuracy when applied to large-scale geospatial data. In this study, we propose a deep learning–geographic information system (GIS) framework for satellite image change detection that integrates remote sensing data with GIS analysis. The framework employs deep learning–based image preprocessing, training data generation, and change detection algorithms, with results mapped onto GIS layers for spatial interpretation. To demonstrate its effectiveness, Sentinel-2 imagery was used for experimental validation. The proposed framework achieved improved accuracy and robustness compared with conventional change detection approaches. The proposed framework achieved high quantitative accuracy and spatial precision compared with conventional normalized difference vegetation index (NDVI) differencing and convolutional neural network (CNN)-based change detection methods. Using one-year Sentinel-2 L2A imagery, our approach reduced false detections by 23% and achieved mean mIoU and precision values of 0.87 and 0.89, outperforming previous CNN models (mIoU = 0.78–0.82, precision=0.84–0.86) These results indicate that the integration of deep learning with GIS provides a practical and scalable solution for environmental monitoring and spatial data analysis.