Nearly 80% of the global population faces severe water stress, but current global assessments cannot reliably track where water is running out and where storage is shrinking, especially in regions with little ground-based data. This project builds a new method to identify global hotspots where both runoff (flowing water in rivers) and storage (water held in glaciers, lakes, reservoirs, and underground) are declining simultaneously. The key innovation is a data-driven model that estimates runoff using satellite measurements of soil moisture, without needing any observed runoff data for calibration—a major advantage in data-sparse regions. The team will combine remote sensing and machine learning to quantify changes in total water storage and its components, then develop multidimensional indices that capture the trade-offs between runoff-limited and storage-limited water threats, which past studies have ignored. If successful, the research will produce the first global map of combined runoff-storage stress hotspots. This matters for infrastructure planners, dam operators, and agricultural policymakers who currently lack reliable information on where water systems are approaching critical thresholds. The findings will help target adaptive strategies—such as reservoir management, irrigation restrictions, or groundwater regulation—to the places that need them most, rather than relying on incomplete or outdated data.
View original technical description
Nearly 80% of the global population faces severe water stress, yet assessment of water use and availability remains in its fancy at the global scale. In data-sparse regions especially, difficulties in accurately estimating changes in runoff and freshwater storage hinder the precise quantification of water stress. Here we develop an interdisciplinary method to identify global hotspots threatened by significant runoff decrease and storage loss using data-driven models and satellite retrievals. Specifically, we build an observation guided data-driven model to estimate runoff using soil moisture dynamics at the global scale. This approach is novel in its independence of any observed runoff data for calibration, particularly showing advantages in data-sparse regions. We combine the strengths of newly developed remote sensing techniques and machine learning models to quantify changes in terrestrial water storage and contributions from glaciers, lakes, reservoirs, and subsurface components. Estimating these changes is vital for interpreting the mechanisms of global storage changes. Moreover, we propose multidimensional indices to quantify runoff-storage stress globally, representing a major advance relative to past studies that neglected the trade-offs between runoff-limited and storage-limited water threats. The global analysis in GLOSTRESS substantially extends novel methods that were initially proposed by the applicant and tested at regional scales. Through targeted collaborations, GLOSTRESS will draw together the interdisciplinary fields of hydrology, remote sensing, environmental science, and water resources. Findings of GLOSTRESS will be valuable not only for the scientific community in understanding hydrologic responses to a changing climate and fast-developing society, but also for policy makers in developing adaptive strategies for water-stressed hotspots.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know