Explaining changes in rainfall-runoff relationships during and after Australia's Millennium Drought: a community perspective

Abstract

Abstract. The Millennium Drought lasted more than a decade, and is notable for causing persistent shifts in the relationship between rainfall and runoff in many south-east Australian catchments. Research to date has successfully characterised where and when shifts occurred and explored relationships with potential drivers, but a convincing physical explanation for observed changes in catchment behaviour is still lacking. Originating from a large multi-disciplinary workshop, this paper presents a range of possible process explanations of flow response, and then evaluates these hypotheses against available evidence. The hypotheses consider climatic forcing, vegetation, soil moisture dynamics, groundwater, and anthropogenic influence. The hypotheses are assessed against evidence both temporally (eg. why was the Millennium Drought different to previous droughts?) and spatially (eg. why did rainfall-runoff relationships shift in some catchments but not in others?). The results point to the unprecedented length of the drought as the primary climatic driver, paired with interrelated groundwater processes, including: declines in groundwater storage, reduced recharge associated with vadose zone expansion, and reduced connection between subsurface and surface water processes. Other causes include increased evaporative demand and interception of runoff by small private dams. Finally, we discuss the need for long-term field monitoring, particularly targeting internal catchment processes and subsurface dynamics. We recommend continued investment in understanding of hydrological shifts, particularly given their relevance to water planning under climate variability and change.

Identifying Causal Interactions Between Groundwater and Streamflow Using Convergent Cross‐Mapping

“…At first glance, this seems inconsistent with the general understanding that GW (baseflow) should contribute more significantly to SF during droughts. However, this long drought seems have strongly affected both GW and SF (Fowler et al., 2022) and produced generally lower SF‐GW connectivity triggered by declining GW levels and soil moisture (Grayson et al., 1997; Hughes et al., 2012; Petrone et al., 2010). This behavior would be consistent with the observation of similar SF hydrograph shapes (Trotter et al., 2022) and lower RR ratios during and after the Millennium Drought (Peterson et al., 2021; Saft, Peel, Western, Perraud, & Zhang, 2016; Saft, Peel, Western, & Zhang, 2016).…”

“…However, the hydrological drought appears to have continued, producing complex RR non‐recovery behavior that challenges the assumption that SF always recovers from drought (Peterson et al., 2021), and there is much speculation as to whether changes in ET and the “long memory” of GW dynamics play a part in the persistence of the hydrological drought (Hughes et al., 2012; Peterson et al., 2021; Petrone et al., 2010; Saft et al., 2015, 2016a, Saft, Peel, Western, & Zhang, 2016; Scaife & Band, 2017). For instance, a recent multi‐disciplinary study suggests the duration of the drought as the main driver of the RR non‐recovery together with interconnected GW processes, namely: declining GW levels and recharge which are associated with unsaturated zone expansion and less interaction between surface and underground water (Fowler et al., 2022). At the same time, the same study highlights other plausible causes, such as increased evaporative demand, and the need for more long‐term field monitoring of internal and subsurface processes to consolidate a physical explanation for the multiyear dynamics.…”

Symptoms of Performance Degradation During Multi‐Annual Drought: A Large‐Sample, Multi‐Model Study

“…The fact that the literature hasn't yet been able to pinpoint the one specific culprit of changes in rainfall-runoff relationships observed during the MD also supports this point (see Fowler et al, 2022, for a review of hypotheses for process understanding of shifts during the MD). Additionally, the fact that metrics associated with the responsiveness of catchments (BFI*) and the timing (r) of flows are less degraded during the drought indicates that despite the models discharging too much flow, the rates of change and timing of discharge are simulated better.…”

“…Note that Stephens et al (2020) and Duethmann et al (2020) both point to poor representation of vegetation dynamics as the culprit for poor DSST performance. Fowler et al (2022) also suggests that hydrological shifts are driven by depletion of storage caused by out-fluxes not reducing during the drought as much as the influxes.…”

“…1997-2009, and discovered that during persistent drought, annual rainfall-runoff relationships shifted significantly in many of the catchments studied; causing reductions in streamflow disproportionate to the meteorological anomaly (Chiew et al, 2014;Potter et al, 2010;Saft et al, 2015). In this context, the annual rainfall-runoff relationship is used to characterise a catchment's response to precipitation and any change in relationship over time can be symptomatic of a modification of a catchment's underlying hydrological behaviour through changes in its underlying processes or their relative prominence, affecting rainfall partitioning (Fowler et al, 2022;Saft et al, 2015). Very similar shifts in rainfall-runoff relationships during prolonged drought were more recently observed also in China (Gao et al, 2016;Tian et al, 2018;Zhang et al, 2018), California (Avanzi et al, 2019), Chile (Alvarez-Garreton et al, 2021) and Europe (Massari et al, 2022).…”