Machine learning (ML) has significantly advanced the understanding of Earth system processes, driven by the rapid growth of high-resolution datasets and data-driven methodologies. In parallel, emerging paradigms including large language models (LLMs), agentic AI and foundation models are beginning to reshape how scientific knowledge is extracted, integrated, and utilized, enabling new forms of reasoning, interpretation, and decision support in Earth science. This session aims to highlight the evolving role of both traditional ML approaches and next-generation AI systems in Earth science, spanning from predictive modeling to knowledge discovery and human–AI interaction. We welcome contributions on radar and satellite observations, data fusion, Earth system modeling and forecasting, natural hazards and extreme events, climate projections, environmental sustainability, as well as explainable AI and LLM-enabled scientific workflows. Studies that advance methodological innovation, improve physical interpretability, or demonstrate generalizable insights are strongly encouraged, particularly in hydrology, precipitation, and related domains.

• A - Atmospheric Sciences