Characteristic mega-basin water storage behavior using GRACE

The NASA GRACE mission now allows hydrologists to study terrestrial water storage variations for the world's largest river basins (>200,000 km2), with monthly time resolution. Because these mega-basins contribute the majority of global runoff, GRACE data are ideally suited for monitoring global water storage variability and classifying differences in basin water storage behavior that are relevant for global climate studies. Here we calculate frequency-domain transfer functions of storage response to precipitation forcing, and then parameterize these transfer functions based on large-scale basin characteristics, such as percent forest cover and basin temperature. This results in a basin-independent relationship between precipitation forcing and storage response as a function of temporal frequency and large-scale basin properties, quantifying fundamental global hydrology relationships that were previously unobservable. Results show that for very large basins, temperature, soil water-holding capacity and percent forest cover are the major controls on relative storage variability, while basin area and mean terrain slope are relatively unimportant. At annual timescales, temperature variability drives storage variability for basins with a mean temperature under 15 deg C, while land surface variables characterize storage variability for warmer basins. At interannual timescales, land surface variables are the largest influence on storage behavior in all basins, with more forested and deeper soiled basins showing reduced response to interannual variability in precipitation forcing. Our results demonstrate the critical importance of forest cover and soil capacity on basin residence times for global-scale studies, and imply that land-atmosphere processes such as precipitation recycling play a critical role in large-basin storage dynamics. The derived empirical relationships were accurate in modeling global-scale water storage anomaly time series for the study basins using only precipitation, average basin temperature, and two land-surface variables, offering the potential for synthesis of basin storage time series beyond the GRACE observational period.