HAPpy Hour Seminar : Geo-web Downscaling Tool (GeoDT) for GRACE: Advancing Groundwater Studies in Data-scarce Regions

Abstract: Groundwater depletion poses a pressing global concern, particularly across working landscapes (e.g., rural, urban, and agricultural) where unsustainable water extraction threatens food production and the socio-ecological system. Insufficient availability of groundwater data creates a significant knowledge gap regarding how climate variability and environmental pressures affect groundwater availability in the working landscapes. Recent progress in remote sensing has facilitated studying of many features of Earth systems, such as vegetation, soil moisture, snow, temperature, and water bodies, by providing data across time and space. The GRACE (Gravity Recovery and Climate Experiment) and its follow-on missions (GRACE-FO) datasets have been widely used to study global groundwater changes. However, the existing GRACE products have still coarse resolution (1 degree to 0.25 degrees ) and lack the spatial detail needed for local decision-making for effective water resource management, particularly across working landscapes.

In this talk, I’ll discuss how machine learning (ML) and artificial intelligence (AI) aided spatial downscaling help to improve the resolution of remote sensing products, narrowing the disparity between community and remote sensing capabilities to provide information and resources that communities need for decision-making. I’ll provide a few examples on downscaling environmental indicators from published case studies. The computationally intensive task of acquiring data from various sources and preparing it for input into ML/AI models makes spatial downscaling difficult for early career students/researchers and water professionals. I’ll talk about how our recently developed tool “Geo-web Downscaling Tool (GeoDT)” can streamline data accessibility from different GRACE products and remote sensing variables, perform downscaling using various ML/AI models at any user-defined resolution (kilometers to meters), and provide visualization of high-resolution data of groundwater storage variations over time. This tool makes groundwater science and data open to researchers and policymakers to get familiar with groundwater changes for actionable decision-making. We are still improving the features of this tool to tailor the community’s needs and plan to host it on NSF NCAR webpage to increase its impact on scientific research in groundwater studies.