The factors driving variability in rainfall stable water isotopes (specifically δ¹⁸O and deuterium excess, d = δ²H - 8 δ¹⁸O) were studied in a 13-year dataset of daily rainfall samples from coastal southwestern Western Australia (SWWA). Backwards dispersion modelling, automatic synoptic type classification, and a statistical model were used to establish causes of variability on a daily scale; and predictions from the model were aggregated to longer temporal scales to discover the cause of variability on multiple timescales. Factors differ between δ¹⁸O and d and differ according to temporal scale. Rainfall intensity, both at the observation site and upwind, was most important for determining δ¹⁸O and this relationship was robust across all time scales (daily, seasonal, and interannual) as well as generalizing to a second observation site. The sensitivity of δ¹⁸O to rainfall intensity makes annual mean values particularly sensitive to the year’s largest events. Projecting the rainfall intensity relationship back through ∼ 100 years of precipitation observations can explain ∼ 0.2-0.4‰ shifts in rainfall δ¹⁸O. Twentieth century speleothem records from the region exhibit signals of a similar magnitude, indicating that rainfall intensity should be taken into account during the interpretation of regional climate archives. For d, humidity during evaporation from the ocean was the most important driver of variability at the daily scale, as well as explaining the seasonal cycle, but source humidity failed to explain the longer-term interannual variability.

In water-limited environments, quantifying the timing and frequency of erratic rainfall recharge events and its climate forcing is of critical importance for groundwater resource management. In temperate semi-arid New South Wales, SE Australia (precipitation: 615 mm/year, pan evaporation: 1679 mm/year), since 2010 we have been using a limestone cave situated at 20 m below land surface, and just above the water table, as a vadose zone observatory of potential recharge approximated by drip rate observations. Complimented since 2018 by a soil moisture probe network and using the VarKarst karst-specialized recharge model, we investigate the climatic, hydrological and karst geological controls on recharge dynamics. We observe nineteen recharge events (07.2010 to 01.2021). They cluster into two periods (1) seven events between 08.2010 and 12.2010 during a La Niña (enhanced spring rainfall is typical in eastern Australia) and (2) seven events between 06.2016 and 10.2016 associated with a negative Indian Ocean Dipole (which is associated with wet winters and springs in southern Australia). Comparison with antecedent rainfall indicates a minimum of 40 mm rainfall over 14-days is required for recharge in winter, and >120 mm rainfall over 14-days in summer. We will use the karst recharge model to simulate the observed recharge events and to quantify the threshold behavior of the soil and vadose zone above the cave. Two recharge events have occurred since the establishment of the soil moisture network (03.05.2020, 29.07.2020). For those, we can analyze the influence of antecedent soil storage on the initiation of recharge and use this understanding for an evaluation of the simulated internal fluxes and storages of karst recharge model. Providing realistic results of both recharge and soil moisture observations, the model can be used as tool to predict the impact of past and future climate changes on groundwater renewal.