Hydrological uncertainties in the modelling of cave drip-water δ18O and the implications for stalagmite palaeoclimate reconstructions

In this paper we review our current understanding of karst drip-water hydrology, emphasising the extent of non-linear and non-stationary process dynamics that render stalagmite palaeoclimate reconstructions using a statistical pseudo-proxy approach difficult to implement. We outline an approach to attribute the uncertainty that arises specifically as a consequence of variable water routing through the overlying soil, epikarst and karst aquifer. This is based upon the development of a monthly lumped parameter karst hydrological model which we use to demonstrate the range of modelled drip-water discharges possible from a single climate input. Refinement of the model, to include precipitation δ18O, enables us to determine the theoretical range in drip-water and stalagmite δ18O for three sites with contrasting climates: northern temperate (NW Scotland), monsoonal (Ethiopia), and Mediterranean (Gibraltar). For actual climate (monthly mean temperature; monthly total precipitation; monthly mean precipitation δ18O), we compare model simulations of karst groundwater storage and drip-water δ18O to demonstrate our ability to model different climate regimes realistically. We also investigate the δ18O variability associated with specific karst water reservoirs that differ in their capacity and drainage mechanisms. δ18O variability is then compared to stalagmite δ18O record from the three regions for the last ∼45 years. We conclude by reviewing the implications of our hydrological model for stalagmite δ18O Quaternary palaeoclimate reconstructions over different timescales and sampling resolutions