Hydrological shifts threaten water resources

Abstract

While droughts are synonymous with lower streamflow, the last 10 years have seen a profusion of examples where drought streamflow was surprisingly low-that is, lower than expected, even given low rainfall. Examples include multi-year droughts in California, USA (Avanzi et al., 2020), Chile (Alvarez-Garreton et al., 2021), China (e.g., Tian et al., 2020 and in the south-west and south-east of Australia (Hughes et al., 2012;Saft et al., 2015). Recently, Peterson et al. (2021) reported that these "shifts" toward lower streamflow can be remarkably persistent, with some watersheds seemingly remaining in a shifted state despite a return to near-average climate conditions. For watersheds remaining in this state, a year of average rainfall today produces less streamflow than it did before the drought (Peterson et al., 2021). Given the potentially global extent of these issues, it is pertinent to consider the policy challenges arising for future water resources planning. A long, slow onset of drier conditions under climate change may induce future amplified shifts which are not accounted for in current water resource projections. Once triggered, the pace of onset of such shifts far exceeds the rapidity of climate change more generally, so climate-change-induced hydrological shifts are a significant threat to water systems over time periods relevant to current operational and planning timescales (10-50 years). Shortcomings of Current Models and MethodsWater resource planners rely on hydrological models to simulate streamflow responses to future climate scenarios. Current modeling techniques produce not only uncertain, but often heavily biased, projections of watershed yield under extended dry conditions. When bias occurs, it is typically toward underestimation of drought risk due to a tendency to downplay hydrologic variability and sensitivity to climatic change (Saft, Peel, Western, Perraud, & Zhang, 2016). Current modeling techniques fail to represent the reported hydrological shifts (Saft, Peel, Western, Perraud, & Zhang, 2016), and a similar failure to anticipate future tipping points could result in "surprise" reductions in actual water availability relative to streamflow projections. These model shortcomings require greater attention from the hydrological community, as well as by scientists and policy makers who create or use hydroclimatic projections. While much attention is given to characterizing climatic uncertainty, hydrological uncertainty is also important and complex (Figure 1). Under climate change, many temperate and sub-tropical regions will likely experience drier conditions and more frequent meteorological droughts (Lehner et al., 2017), and a key aspect of improving future water security is understanding how these changes (characterized by precipitation) might translate into impacts on rivers. These impacts include changes in streamflow (yield) as well as