Analyzing the land cover temporal changes is crucial for handling urban growth, agricultural development, forest conservation, and water resources. However, conventional land cover classification methods are often tedious, effort-intensive, and unreliable without automation. In order to tackle these shortcomings, we propose a deep learning enabled method for automated land cover classification using multispectral satellite imagery. Precisely, we implement architecture with a ResNet50 encoder to classify images from the EuroSAT dataset into 10 distinct land cover types, including annual crop, forest, residential, river, and Sea Lake. The EuroSAT dataset, derivative from Sentinel-2 imagery, provides abundant multispectral data; we exploit five key spectral bands (B2–B6) to construct model inputs. The model is achieved a training accuracy of 96% and a validation accuracy of 93%, indicating strong generalization. This method significantly reduces the need for manual surveys and expedites efficient, scalable land cover mapping. The results support applications in sustainable land use planning and environmental policy. Future work can focus on integrating larger datasets and advanced data augmentation strategies to further enhance the classification accuracy.