Remote sensing and Earth observation are advancing through integration of multi-source satellite and UAV data with machine learning and large language models. Physical machine learning—combining physics knowledge with data-driven approaches through physics-informed constraints and neural operators—enhances consistency and generalization. Geospatial foundation models and multimodal language systems enable knowledge transfer across sensors and support open-vocabulary perception and reasoning.
This special issue seeks original research on physics-informed learning, geospatial foundation models, and language-based intelligence for remote sensing applications. Submissions should advance methodological approaches, establish benchmarks, or demonstrate applications that improve accuracy, generalization, interpretability, and operational value across environmental monitoring, resource management, and disaster response. Studies combining these complementary approaches are particularly encouraged.