Monitoring changes in freshwater availability is critical for human society and sustainable economic development. To identify regions experiencing secular change in their water resources, many studies compute linear trends in the total water storage (TWS) anomaly derived from the Gravity Recovery and Climate Experiment (GRACE) mission data. Such analyses suggest that several major water systems are under stress (Rodell et al 2009 Nature 460 999–1002; Long et al 2013 Geophys. Res. Lett. 40 3395–401; Richey et al 2015 Water Resour. Res. 51 5217–38; Voss et al 2013 Water Resour. Res. 49 904–14; Famiglietti 2014 Nat. Clim. Change. 4 945–8; Rodell et al 2018 Nature 557 651–9). TWS varies in space and time due to low frequency natural variability, anthropogenic intervention, and climate-change (Hamlington et al 2017 Sci. Rep. 7 995; Nerem et al 2018 Proc. Natl Acad. Sci.). Therefore, linear trends from a short time series can only be interpreted in a meaningful way after accounting for natural spatiotemporal variability in TWS (Paolo et al 2015 Science 348 327–31; Edward 2012 Geophys. Res. Lett. 39 L01702). In this study, we first show that GRACE TWS trends from a short time series cannot determine conclusively if an observed change is unprecedented or severe. To address this limitation, we develop a novel metric, trend to variability ratio (TVR), that assesses the severity of TWS trends observed by GRACE from 2003 to 2015 relative to the multi-decadal climate-driven variability. We demonstrate that the TVR combined with the trend provides a more informative and complete assessment of water storage change. We show that similar trends imply markedly different severity of TWS change, depending on location. Currently more than 3.2 billion people are living in regions facing severe water storage depletion w.r.t. past decades. Furthermore, nearly 36% of hydrological catchments losing water in the last decade have suffered from unprecedented loss. Inferences from this study can better inform water resource management.
- linear trends
- spatiotemporal variability