Ecosystem water use plays a large role in determining surface water availability, and thus streamflow, and society's water resources. Water use and streamflow are highly sensitive to multiple climate extremes, many of which are co-occurring and compounding events. It is the co-occurrence of extremes that poses the biggest risk to ecosystems and society, as co-occurring climate extremes amplify impacts due to multiple nonlinear interactions. Coastal regions of the western US exhibit strong spatial gradients in water resources and provide an excellent testbed to understand the influence of compound and sequential extreme events on ecosystem water use and the resulting effects on streamflow and water relations. In tandem, dense observation networks such as AmeriFlux and the USGS stream gage network provide an opportunity to examine the impact of extremes on coastal water resources, as we move to no-analogue environments in the coastal regions of the future. This project will examine the impact of extreme drought, heat, and wildfire on water relations of the coastal western US, integrating machine learning, Earth System Models, and emulators with distributed sensor networks to do so. The strategic combination of novel observationally informed techniques within a model benchmarking framework, combining both the water and carbon cycles along with associated uncertainties, will greatly improve our understanding of the relationship between coastal ecosystem function and current and future water resources.