Separating isotopic impacts of karst and in-cave 2 processes from climate variability using an integrated 3 speleothem isotope-enabled forward model 4

Pauline C. Treble, Mukhlis Mah , Alan Griffiths, Andy Baker , Michael 5 Deininger, Bryce F. J. Kelly , Denis Scholz, Stuart I. Hankin 6 7 ANSTO, Lucas Heights, NSW 2234, Australia 8 Connected Waters Initiative Research Centre, UNSW Sydney, Sydney 2052, Australia 9 School of Biological, Earth and Environmental Sciences, UNSW Sydney, Sydney 2052, Australia 10 Institute for Geosciences, University of Mainz, Johann-Joachim-Becher-Weg 21, 55128 Mainz, 11 Germany 12


In-Cave Processes (No Karst Processes)
Model of stalagmite growth and δ 13 C values on the stalagmite surface solution.
[ Stoll et al., 2012] Model of the temporal isotopic (δ 18 O and δ 13 C) evolution of DIC in a thin film precipitating calcite.

Soil and In-Cave Processes (No Karst Processes)
ODSM Model of stalagmite δ 18 O values from climatic input, soil mixing and vegetation effects. Soil water was modelled straight to in-cave and temperature dependent fractionation applied. [Wackerbarth, 2012;Wackerbarth et al., 2010;Wackerbarth et al., 2012] Model of δ 13 C and δ 18 O values in soil (against soil pCO 2 ) and in-cave isotope fractionation processes.
[ Dreybrodt and Scholz, 2011] CaveCalc PHREEQC-based model of soil, bedrock and in-cave processes including isotopes and trace elements Owen et al. (2018)

Karst Processes (No In-Cave Processes)
Climatically fed single reservoir model with fracture flow for high magnitude rainfall and diffuse flow for low magnitude rainfall.
[ Baker et al., 2010;Nagra et al., 2016] Two layer reservoir dripwater δ 18 O model based on climatic input, with stores modelled as steady state.
[ Truebe et al., 2010] KarstFor Three (or four) layer reservoir model with soil evaporation, monthly water balance, overflow and underflow based on climatic input. Dripwater δ 18 O values include temperature dependant isotope fractionation.

148
See Table 2 for a summary of model inputs and parameters.

150
The residence time of the surface water layer on the stalagmite controls the degree of the resulting 151 isotopic disequilibrium [Deininger et al., 2012;Scholz et al., 2009]. As this residence time is dictated 152 by the drip interval, which in turn is dependent on karst hydrology, modelling of the drip interval from 153 the karst is necessary for application of ISOLUTION. We model drip-infiltration as gravity fed, with a 154 linear response to the volume present in the karst store. Users specify the drip rate when the store is 155 empty, ! , and full, ! , and the model calculates the instantaneous drip rate, according to where ksize is the capacity of the karst store directly supplying the drip (mm) and kstor is the current 157 level of the karst store (mm; blue line in the stores shown in Fig. 1). The modelled drip interval, , is This allows for a choice of ! and ! where dripping stops before the store empties completely and may be affected by factors such as land use change, shading, fire, and rapid climate change 167 [Domínguez-Villar et al., 2015;Nagra et al., 2016]. To deal with this, surface-cave temperature 168 coupling is implemented in Karstolution with a site-specific difference between the ground surface and 169 cave air temperature (Δ !!! ). While maintaining Δ !!! , cave temperature varies using a user-defined 170 moving average of the surface temperature (36-months in this study). A useful guide for temperature-171 depth penetration at different time periods is, for instance, presented in Fig. 9 of Rau et al. [2015].

189
The two parameter Weibull distribution is represented as The two parameters, k and λ, represent shape and scale, respectively (Fig. 2). The Weibull function is implemented over the domain 0 ≤ x ≤ 2, divided over an adjustable-length mixing window. At each 192 model step, k_diffuse determines the diffuse flow leaving the epikarst (see Table 2). This, along with where is the pCO 2 of water droplets, app is the apparent equilibrium pCO 2 of cave air (1/ 0.8 times 214 the cave air pCO 2 ), is the water film thickness (set to 0.01 cm), Δ is the drop interval, α is a rate 3 Site description

239
• relative humidity with a Vaisala HMP155 with Humicap 180RC and sensor warming enabled 240 to negate saturation of the sensor at high humidity (accuracy ±1.8%); • cave ventilation with a Gill Windsonic (±2%); and • cave air pCO 2 with a Viasala GMP252 with measurement range 0-10,000 ppm (accuracy

288
A sensitivity analysis was performed to determine which cave parameters potentially drive stalagmite 289 δ 18 O variability at Golgotha Cave. Figure 6 shows a reference case based on Site 1A, along with 290 perturbations to cave parameters (Fig. 4). It demonstrated that the isotopic impact of temperature 291 should be considered once mean cave temperature variability exceeds ±2°C (i.e., greater than that 292 expected for the Holocene).

293
In terms of the potential isotopic impact of disequilibrium processes, stalagmite δ 18 O appears to have 294 some sensitivity to drip interval over the observed range for Golgotha Cave (Fig. 6b)