Climate extremes such as drought, flood, heat wave, heavy precipitation, and wildfire are likely to be more severe, frequent, and intensified in the future, posing great risks to agriculture and urban systems. Advances in data sciences (e.g. machine learning, deep learning, reinforcement learning, causal inference) and emerging earth data from novel sources can be exploited to provide a predictive understanding of different individual or compound climate extremes as well as their impacts across scales. This session welcomes submissions of process-based and data-driven research including but are not limited to: approaches for better monitoring, forecasting, or analyzing causes of climate extremes, downscaling and bias correction, diagnostic evaluations, or uncertainty analysis for modeling, predictions, or impact assessment of climate extremes, inference and decision-making under uncertainty, adaptations or mitigations of agricultural and urban systems to climate extremes, and engagement of stakeholders to develop and implement data-informed best management practices or decision support tools.