The field of remote sensing is rapidly advancing with the integration of deep learning techniques, enabling more accurate and efficient analysis of Earth observation data. Recent developments have focused on improving the resolution and accuracy of land cover classification, object detection, and scene categorization. The use of transformer-based architectures and self-supervised learning methods has shown promising results in various applications, including glacier calving front extraction and landslide detection. Additionally, the development of large-scale datasets and pre-trained models has facilitated the advancement of remote sensing research. Notably, the introduction of physics-driven deep learning models has improved the forecasting of complex phenomena such as storm surges and iceberg drift. Overall, the field is moving towards more accurate, efficient, and scalable analysis of remote sensing data, with significant implications for environmental monitoring, disaster risk management, and sustainable land-use planning. Noteworthy papers include: Storm Surge in Color, which introduces a novel approach to storm surge forecasting using RGB-encoded physics-aware deep learning; IDRIFTNET, which proposes a physics-driven spatiotemporal deep learning model for iceberg drift forecasting; and SelvaBox, which presents a large-scale dataset for tropical tree crown detection. These papers demonstrate the innovative applications of deep learning and remote sensing techniques in advancing our understanding of the Earth's systems and addressing pressing environmental challenges.