A Mechanism-Coupled Split Window Network for Medium- to High-Resolution Land Surface Temperature Retrieval
Researchers have developed a new neural network framework called PCD-Net to improve the retrieval of land surface temperature (LST) from satellite data. Traditional methods struggle with complex atmospheric conditions and diverse land covers, while purely data-driven models lack generalizability. PCD-Net addresses this by reformulating the retrieval as a dynamic learning problem for physical component coefficients, explicitly modeling the relationships between land surface emissivity, atmospheric water vapor, and temperature differences. AI
IMPACT This new AI framework could lead to more accurate climate modeling and environmental monitoring by improving land surface temperature data.