Insights into recharge processes and speleothem proxy archives from long-term monitoring networks of cave drip water hydrology   Andy Baker1, Pauline C. Treble2, 1, Andreas Hartmann3, 1, Mark O. Cuthbert4, 1, Monika Markowska5, Romane Berthelin3, Carol V Tadros2, 1, Matthias Leopold6, Stuart Hankin2 and the KSS Cave Monitoring Team7   1Connected Waters Initiative, UNSW Sydney, Australia Contact Andy at a.baker@unsw.edu.au 2ANSTO, Lucas Heights, Australia 3University of Freiburg, Germany 4Earth and Ocean Sciences, Cardiff University, UK 5 Max Planck Institute for Chemistry, Mainz, Germany 6School of Agriculture and Environment, University of Western Australia, Perth, Australia 7 Kempsey Speleological Society What is the rainfall amount needed to generate recharge? Long-term monitoring of drip hydrology using networks of automated loggers can be used to compare recharge events to antecedent rainfall using both observational data and modelling. Multiple drip loggers deployed in a chamber in Cathedral Cave, Wellington, NSW (photo: Ian Eddison) Upper Macleay Lower Macleay Kempsey AWS Macleay River Kempsey Location of the Macleay Karst region. Inset box shows the region shown: Upper and Lower Macleay cave monitoring sites (circles) and the nearest weather station (Kempsey automated weather station). Image is from Google Earth, used with permission. Mean annual temperature (MAT) of 18.6 °C and total annual precipitation (P) of 1218 mm. Annual P/PET varied between 0.62 (2016 CE) to 1.13 (2015 CE). Example 1: Macleay Karst, New South Wales, Australia (in revision, Journal of Hydrology, 10.1016/j.jhydrol.2020.125001) Fourteen continuous drip discharge time series from 2014 CE onwards, in a karstified, fractured limestone, from six caves of varying depths and vegetation cover in the sub-tropical humid Macleay Valley of the mid-north coast of New South Wales, Australia. Upper Macleay Lower Macleay Kempsey AWS Macleay River Kempsey (a) Daily precipitation from Lower Macleay (black) and Upper Macleay (red). (b) Daily AET (c) Daily cumulative water balance. (d) Event number. (e) All drip logger data for the Lower Macleay (red) and Upper Macleay (black). We observe thirty-one cave drip water recharge events over a five-year monitoring period. Monitoring is on-going. Lower Macleay. The monitored caves are situated c. 40 m above the valley floor in this isolated karst hill. The source of rainfall recharge is only from this hill. Photo taken during the 2019 drought and bushfires. The entrance to one of our monitoring sites in the subtropical rainforest. Carter Cave, Upper Macleay We used a simple water balance model (below) to quantify soil and epikarst water storage volumes and to test hypotheses of the hydrological controls. Driven by P and AET, two model parameters, soil and epikarst water storage capacity (V) and diffuse drainage (D) were calibrated to obtain the optimised solution such that overflow R occurred at the same time as observed recharge events. Modelled storage capacity (V) of about 65 mm (Lower Macleay) and 80 mm (Upper Macleay) confirm a correspondence between observed weekly precipitation thresholds and soil and epikarst capacities (see Table, part B). Comparison to antecedent precipitation (see Table, part a) demonstrates a median observed recharge threshold of 76 mm / week precipitation (Lower Macleay) and 79 mm / week precipitation (Upper Macleay), with lower precipitation thresholds (down to 30 mm / week) possible. Model code at https://github.com/KarstHub/Simple-Water-Budget-Model. Credit: Mirjam Scheller (a) Recharge thresholds (mm) Lower Macleay Upper Macleay   Antecedent precipitation Antecedent precipitation 7-day 14-day 21-day 7-day 14-day 21-day MEAN 85.3 111.2 130.2 MEAN 97.6 124.5 153.4 MEDIAN 76.2 98.6 127.1 MEDIAN 79.4 108.0 149.8 MIN 33.3 43.1 43.4 MIN 30.1 32.6 47.9                   (b) Water balance model     Lower Macleay Upper Macleay Overflow Capacity (mm) 65 80 Drainage (mm) 4.8 0.5 Number of observed events 14 (+3 low volume events, with indistinct hydrographs) 27 Number of observed events correctly simulated 7 (+3 low volume events with indistinct hydrographs) 19 Number of observed events not simulated 7 9 Number of simulated events not observed 0 3 Example 2: Wellington, New South Wales, Australia (Markowska et al., 2020; Cuthbert et al., 2014, and unpublished data) At Cathedral Cave, NSW, with a temperate semi-arid climate, we have almost ten years of drip hydrology monitoring from a network of drip loggers. Mean total annual precipitation is 629 mm and mean annual temperature is 18.2 °C. P/PET = 0.45 (Baker et al., 2019). The time series graphs on this slide show the normalised hydrology for the eighteen most reliable loggers. The 2019 drought year is evident. We now have enough recharge events (n=21) to determine the antecedent rainfall volume needed for drip recharge, using the methodology presented previously. This will be completed through a 2020 UNSW Honours research project. Photo: Ian Eddison We previously applied a soil moisture balance - karst model (schematic below) to the drip recharge data for 2010-2014 (Cuthbert et al. 2014). The best fit between observed recharge events and modelled recharge (Rch) was determined. Modlelling suggested a maximum soil moisture deficit of 87 mm is needed to be overcome prior to cave drip water recharge. Matlab model code available at: https://doi.org/10.6084/m9.figshare.10282868 This isotope enabled model was then used to compare modelled and measured d18O for two modern stalagmites, WB and WC (Markowska et al., 2020). The isotope model only includes a parameterised rainfall amount effect and isotope fractionation due to epikarst evaporation. For stalagmite WC, almost all major peaks and troughs could be replicated within the age uncertainty, suggesting that epikarst evaporation provides a good explanation for the variability in WC d18Ospel. WC can be explained by a simple single-reservoir model, which is very sensitive to karst evaporation and changes in water storage volumes, controlled by alternating periods of recharge and decline. WB cannot be explained by a simple single-reservoir model and likely is fed by multiple reservoirs. How unique is a drip? What is the heterogeneity in drip rate between sites within a cave, and between caves? Harrie Wood Cave South Glory Cave Long-term monitoring of two caves at the montane Yarrangobilly Caves, NSW by ANSTO and UNSW teams can be used to quantify variability in drip hydrology response. Published in Markowska et al. (2015) Published in Coleborn et al. (2016) At Yarrangobilly, mean annual temperature = 11.1 °C. Mean total annual precipitation = 1177 mm. P/PET = 1.07 Up to 0.3‰ Up to 0.9‰ iSOLUTION modelled fractionation The current dataset comprises over four million unique drip rate measurements over eight years, from twenty drip sites, from two caves (red – South Glory Cave; Black – Harrie Wood Cave). Cumulative frequency drip rate distributions for each of the twenty drip sites is shown on the left. All drip rates are from the same time period and same hydroclimate forcing. This graph reveals the natural varaibility of drip discharge. Two drip rates (180 s and 750 s per drip) are shown as examples. Modelled drip water oxygen isotope fraction using iSOLUTION is shown in blue. The graph below shows that drip rates slower than 180 s per drip are experienced at these monitored drip sites for 11.8% of the time (75% quartile). Box plots of: (a) drip rate; (b) drip water δ18O values, δ18Odw (VSMOW) and stalagmite δ18O values, δ18Ostal (VPDB); from sites HW1, HW2 and HW38b in Harrie Wood Cave, south-eastern Australia. Grey shading in (b) represents the box plot of calcite-equivalent δ18Odw (VPDB) values for comparison with calcite δ18O (VPDB) measurements. ISOLUTION modelled evolution of speleothem calcite δ18O values with varying drip intervals. The solid line represents the modelled evolution from the mean HW1 δ18O drip water value and the dashed lines correspond to the 90% confidence interval. Long term monitoring of both cave drip water oxygen isotope composition and drip rates therefore allows the quantification of the extent of speleothem oxygen isotope variability due to fractionation associated with changes in drip rate. ANSTO long-term monitoring site, Yarrangobilly Caves, NSW, Australia. Carol Tadros and Pauline Treble, unpublished data (Tadros et al. in prep) At Golgotha Cave, Western Australia, Electrical Resistivity Tomography (ERT) measurements used an array of electrodes on the surface above the cave (upper figure). Differences in resistivity can be linked to the water saturation of the rock profile at this site where flow through both primary porosity and fractures is possible. In the left-hand figure, low resistivity (blue) = water-filled cavities / porous damp rock. High resistivity (maroon) = dry/solid rock. ‘Time lapse’ repeat surveys such as this can image water movement. Note the connected saturated rock from surface to depth in winter, and a late summer disconnection at the surface (February 2017). Long-term monitoring can be combined with time-lapse electrical resistivity tomography (ERT) to observe seasonal patterns in unsaturated zone water movement to better interpret the speleothem oxygen isotope archive. 11 ERT data corresponds with the strongly seasonal water balance (upper left figure) in this region of Mediterranean climate. Long-term monitoring identifies an inverse relationship between drip rate and drip water d18O (middle left figures), due to the influence of fracture flow leading to lower drip water d18O. Fractures activate in wetter years e.g. 2013, and local rainfall water isotope data (lower left figure) shows that wetter years typically lower rainfall d18O due to the “amount effect” for region. Differences between speleothem d18O time series and ‘farmed calcite’ (lower figure) mirror that of drip water, and are explicable by the heterogeneity in karst hydrology. 12 Conclusions We demonstrate the utility of long-term, spatially dense monitoring of drip water hydrology, combined with hydrological modelling, measurement of drip water isotope composition, and time-lapse electrical resistivity tomography. Networks of automated drip loggers permits the quantification of the recharge bias for speleothem forming drip waters. The focus here is on Australian, water limited environments (E >= P). They also permit the quantification of the heterogeneity of speleothem forming drip water discharge. The focus here is on quantifying the hydrological controls on the within-cave variability in drip water oxygen isotopic composition, to better quantify uncertainties in the speleothem oxygen isotope archive. References Baker, A. Berthelin, R., Cuthbert, M.O., Treble, P.C., Hartmann, A. and the KSS Cave Studies Team. Rainfall recharge thresholds in a subtropical climate determined using a regional cave drip water monitoring network. J. Hydrology, in revision. Baker, A., Hartmann, A., Duan, W., Hankin, S., Comas-Bru, L., Cuthbert, M.O., Treble, P.C., Banner, J., Genty, D., Baldini, L., Bartolomé, M., Moreno, A., and Pérez-Mejías, C., 2019. Global distribution and controls on cave drip water oxygen isotope composition. Nature Communications, 10, Article number: 2984 Coleborn, K., Rau, G.C, Cuthbert, M.O., Baker, A and Navarre, O., 2016. Solar forced diurnal regulation of cave drip rates via phreatophyte evapotranspiration. Hydrology and Earth System Sciences, 20, 4439-4455 Cuthbert, M.O., Baker, A., Jex, C.N., Graham, P.W., Treble, P., Andersen, M.S and Acworth, R.I., 2014 Drip water isotopes in semi-arid karst: implications for speleothem paleoclimatology. Earth and Planetary Science Letters, 395, 194-204 Markowska, M., Andersen, M.S., Treble, P.C., Baker, A., Tadros, C., Hankin, S., and Jex, C.N., 2015. Unsaturated zone hydrology and cave drip discharge water response: Implications for speleothem palaeoclimate record variability. J. Hydrology, 529, 662-675. Markowska, M., Cuthbert, M.O., Baker, A., Treble, P.C., Andersen, M.S., Adler, L., Griffiths, A., and Frisia, S., 2020. Modern speleothem oxygen isotope hydroclimate records in water-limited SE Australia. Geochimica et Cosmochimica Acta, 270, 431-448. image1.png image2.png image3.png image4.jpg image5.png image6.jpg image7.png image8.jpg image9.png image10.png image11.png image12.png image13.tif image14.tif image15.png image16.png image17.png image18.emf 1/07/20101/07/20111/07/20121/07/20131/07/20141/07/20151/07/20161/07/20171/07/20181/07/2019 100 80 60 40 20 0 Number of dry drips (0 = all wet, 100 = all dry) image19.emf 1/07/20101/07/20111/07/20121/07/20131/07/20141/07/20151/07/20161/07/20171/07/20181/07/2019 0 10 20 30 40 50 60 70 80 90 100 Dripwater recharge (max = 100, min = 0) image20.png image21.jpg image22.tiff image23.emf -7 -6 -5 -4 -3 -2 1930194019501960197019801990200020102020 1930194019501960197019801990200020102020 -5 -5 -4 -4 -3 -3 Stalagmite WB modelled Year (CE) d 18 O ( ‰ VPDB) Stalagmite WC modelled image24.tif image25.emf 1/01/20111/01/20121/01/20131/01/20141/01/20151/01/20161/01/20171/01/20181/01/20191/01/2020 0 50 100 150 200 250 300 350 400 Drips / minute image26.png image27.emf 0 60 120180240300360420480540600660720780840900960 0 20 40 60 80 100 Cumulative Frequency (%) Time between drips (s) image28.emf 0.3 11.8 0.0 1.5 0 20 40 60 80 100 Percentage of time Drip rate > 180 s Drip rate > 750 s image29.png image30.png image31.png image32.png image33.png image34.png image35.png image36.png image37.png image38.png /docProps/thumbnail.jpeg