GRACE storage-runoff hystereses reveal the dynamics of regional watersheds
- 1Oak Ridge Institute for Science and Technology, c/o National Health and Environmental Effects Research Laboratory, US Environmental Protection Agency, 200 SW 35th Street, Corvallis, OR, 97333, USA
- 2College of Earth, Ocean, and Atmospheric Sciences, Oregon State University, Corvallis, OR, 97331-5503, USA
- 3National Health and Environmental Effects Research Laboratory, US Environmental Protection Agency, 200 SW 35th Street, Corvallis, OR, 97333, USA
- 4Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, 91109, USA
- 5School of Environment, Natural Resources and Geography, Bangor University, Bangor, LL57 2UW, Wales, UK
- 6Centro de Estudios Avanzados en Zonas Áridas, Universidad de La Serena, Raul Bitran 1305, La Serena, Chile
Abstract. We characterize how regional watersheds function as simple, dynamic systems through a series of hysteresis loops using measurements from NASA's Gravity Recovery and Climate Experiment (GRACE) satellites. These loops illustrate the temporal relationship between runoff and terrestrial water storage in three regional-scale watersheds (> 150 000 km2) of the Columbia River Basin, USA and Canada. The shape and size of the hysteresis loops are controlled by the climate, topography, and geology of the watershed. The direction of the hystereses for the GRACE signals moves in opposite directions from the isolated groundwater hystereses. The subsurface water (soil moisture and groundwater) hystereses more closely resemble the storage-runoff relationship of a soil matrix. While the physical processes underlying these hystereses are inherently complex, the vertical integration of terrestrial water in the GRACE signal encapsulates the processes that govern the non-linear function of regional-scale watersheds. We use this process-based understanding to test how GRACE data can be applied prognostically to predict seasonal runoff (mean Nash-Sutcliffe Efficiency of 0.91) and monthly runoff during the low flow/high demand month of August (mean Nash-Sutcliffe Efficiency of 0.77) in all three watersheds. The global nature of GRACE data allows this same methodology to be applied in other regional-scale studies, and could be particularly useful in regions with minimal data and in trans-boundary watersheds.