Key points are not available for this paper at this time.
Every year, natural climate variability leads to droughts and floods which have significant impacts for ecosystems and societies. Water reservoirs like soil moisture, lakes, and groundwater act as natural buffers and balance these fluctuations by providing water supply during dry conditions and by storing water surplus after rain and snow events. Such natural fluctuations unfold over time scales that can reach several decades, making it challenging to assess the extent to which trends in water reservoirs observed over the recent past are caused by anthropogenic modifications. Such modifications can themselves be further partitioned into different terms. For instance, one can contrast the contribution of regional land and water management on the one hand, and the contribution of climate change on the other. Another frequent framework is to causally relate changes in water storage to individual changes in precipitation, evapotranspiration, and runoff. In this contribution, we review the strengths and weaknesses of recent approaches used to causally attribute observed as well as projected changes in water availability. Ensembles of model simulations and factorial experiments typically represent a powerful way of assessing individual responses to drivers and developing a plausible and mechanistic understanding. However, contradictions also quickly emerge between global hydrological model simulations, which typically represent water reservoirs and water management more thoroughly, and Earth system (climate) model simulations, which include biogeochemical effects, like CO2 fertilization, that are typically neglected by hydrological models. We will show that these two incomplete modeling worlds can be reconciled with large-scale satellite observations in only a few regions, while very large uncertainties remain in other parts of the world and in particular over tropical areas.
Humphrey et al. (Mon,) studied this question.