New modelling paradigm linking groundwater, surface water and rainfall-runoff relationship shifts under multi-year drought

While there is little opposition to the idea that groundwater can play a central role in rainfall-runoff response, there is little consensus on how this should be modelled. Here, we present modelling supporting a recently advanced hypothesis that links groundwater-surface water interactions with observed shifts in the relationship between rainfall and runoff in south-east Australia.  While many approaches assume a direct and simplistic relationship between groundwater head and baseflow, evidence arising from a multi-year drought in Australia challenges traditional notions. GRACE and bore data during the Millennium Drought (1997-2010) show multi-year declines in groundwater storage, of such severity that we might expect the baseflow to cease, giving a flashier regime.  In reality, the shape of the hydrograph is mostly unchanged, but other changes abound: a year of given rainfall generates less runoff today than it did pre-drought (ie. shift in rainfall-runoff relationship). In other words, during and after the drought we see a hydrograph of similar shape to before, but diminished. While many Australian hydrologists are convinced that groundwater played a key role in this behaviour, it is unclear how these observations can be explained by existing hypotheses or modelling methods for groundwater surface-water interaction, and new paradigms are required. The hypothesis explored here is that these observations can be explained by leaky bedrock in headwater catchments, which facilitates gradual groundwater export from upslope areas to downslope areas (within the same catchment).  Upslope areas subject to groundwater decline then see groundwater-surface water decoupling and reduced runoff. The hypothesised leakage is slow enough to go unnoticed during wetter periods, but in drier periods recharge may be too low to balance the export, leading to reduced groundwater levels and groundwater surface-water decoupling. When wetter conditions resume, the groundwater deficits may take a while to be replenished, delaying recovery of rainfall-runoff relationships (as observed in Australia).  In downslope areas, the drained water may contribute to streamflow, but may also be lost to evaporation and transpiration, particularly in drier catchments with flatter valley bottoms of alluvium or colluvium. In such catchments, the net effect of these processes is to allow groundwater originating from upslope to supplement evaporative budgets downslope rather than increasing streamflow. We advance this hypothesis, firstly by presenting evidence of its applicability in south-east Australia; and secondly by building and testing improved numerical models that incorporate a simplified representation of these processes. Modelling results show improved performance when tested across several catchments affected by rainfall-runoff relationship changes, and improved realism such as multi-year declines in simulated groundwater storage, consistent with observations.  These results suggest a promising avenue for further research relevant to a variety of water resource applications including climate change impact assessment.