The field of ecohydrological modeling is moving towards the integration of process-based models with machine learning and artificial intelligence techniques. This shift is driven by the need for more accurate and efficient modeling of complex environmental systems, particularly in the context of climate change and human pressures. Recent developments have focused on the use of knowledge distillation, graph neural networks, and vision language models to improve the accuracy and interpretability of ecohydrological models. Notably, these advancements have also been applied to disaster response and recovery, including wildfire damage assessment, flood prediction, and urban flood depth estimation.

Some noteworthy papers in this area include: Automated Wildfire Damage Assessment from Multi view Ground level Imagery Via Vision Language Models, which demonstrates the efficacy of vision language models in synthesizing information from multiple perspectives to identify nuanced damage. RiverScope: High-Resolution River Masking Dataset, which provides a valuable resource for fine-scale and multi-sensor hydrological modeling, supporting climate adaptation and sustainable water management. HydroGAT: Distributed Heterogeneous Graph Attention Transformer for Spatiotemporal Flood Prediction, which achieves higher accuracy and improved interpretability in flood prediction through the use of a heterogeneous basin graph and adaptive learning of local temporal importance. MCANet: A Multi-Scale Class-Specific Attention Network for Multi-Label Post-Hurricane Damage Assessment using UAV Imagery, which learns multi-scale representations and adaptively attends to spatially relevant regions for each damage category, achieving a mean average precision of 91.75%. FloodVision: Urban Flood Depth Estimation Using Foundation Vision-Language Models and Domain Knowledge Graph, which combines the semantic reasoning abilities of a foundation vision-language model with a structured domain knowledge graph to enable accurate depth estimation that can generalize across diverse flood scenarios. An Explainable Deep Neural Network with Frequency-Aware Channel and Spatial Refinement for Flood Prediction in Sustainable Cities, which integrates novel components to redefine urban flood classification through advanced deep-learning techniques, achieving state-of-the-art F1-scores on benchmark datasets.