The field of spatiotemporal modeling and change detection is witnessing significant advancements, driven by the development of innovative deep learning architectures and techniques. Researchers are focusing on improving the accuracy and robustness of models in capturing complex spatial and temporal dependencies, particularly in applications such as water resource management, remote sensing, and environmental monitoring. Notable developments include the integration of attention mechanisms, diffusion-based methods, and probabilistic techniques to enhance model performance and provide uncertainty estimates. These advancements have the potential to support informed decision-making in various domains. Noteworthy papers include: AdaTrip, which introduces an adaptive graph learning framework for multi-reservoir inflow forecasting. DiffRegCD, which presents a unified framework for dense registration and change detection, leveraging diffusion features and classification-based correspondence. ForeSWE, which proposes a probabilistic spatio-temporal forecasting model for snow-water equivalent prediction, incorporating attention mechanisms and Gaussian process modules. USF-Net, which develops a unified spatiotemporal fusion network for ground-based remote sensing cloud image sequence extrapolation, combining adaptive large-kernel convolutions and low-complexity attention mechanisms.