Global database of karst springs and cave drips opens a new window on hidden groundwater

Deep beneath some of the world’s most iconic landscapes, water moves through a labyrinth of dissolved limestone channels, fractures, and porous rock at speeds that can range from a slow trickle to a torrential rush. Karst aquifers of this kind supply drinking water to roughly nine percent of the global population, and carbonate rock terrains are home to about 16.5 percent of all people. Yet despite their importance, these aquifers remain among the most difficult water systems on Earth to understand, because their internal plumbing is extraordinarily heterogeneous and behaves in strongly nonlinear ways. A new international effort now promises to change that, with the release of the first global database of stable isotope records from karst springs and cave drip water.

The database, called WoKaS-Iso, was compiled by a team of more than fifty researchers led by Yining Zang and Andreas Hartmann of the Technical University of Dresden, together with Pauline C. Treble of Australia’s Nuclear Science and Technology Organisation. Published in the journal Earth System Science Data, it brings together 1007 time series of oxygen-18 and deuterium measurements drawn from 241 karst springs and 74 caves across China, the United States, Europe, the Middle East, and Australia. The records span nearly six decades, from 1967 to 2024, with the densest coverage during the 2010s and continuing into the early 2020s.

The scientific power of the database lies in what the isotopes reveal. Oxygen-18 and deuterium are heavy, naturally occurring variants of the atoms that make up every water molecule, and their ratios to the common forms, expressed in delta notation relative to Vienna Standard Mean Ocean Water, act as fingerprints of where water fell as rain and how it travelled underground. Because these tracers are part of the water itself, they can identify recharge sources, delineate subsurface flow paths, estimate how long water takes to move through an aquifer, and quantify mixing between fast conduit flow and slow storage in the rock matrix. Spring isotope signatures integrate responses from both the saturated and unsaturated zones, providing information that discharge measurements alone cannot capture.

Cave drip water offers a complementary and equally valuable window. Drips feeding stalagmites and other speleothems pass through the epikarst, the weathered upper layer of rock, and their isotope composition is directly linked to the climate signals preserved in the cave formations that grow from them. Long-term monitoring of rainfall and drip water is therefore crucial for distinguishing local from regional climatic influences on cave isotope variability, and for refining the interpretation of speleothem records used to reconstruct past climates. Recent work has shown how powerful this approach can be: in southwestern Australia, combining drip water and speleothem isotope data with modelled soil moisture revealed that regional drying has disrupted rainfall recharge to shallow karst aquifers in a way unprecedented over the past 800 years.

Until now, however, isotope observations from karst environments have been scattered across individual studies, theses, and technical reports, while global isotope databases such as the Global Network of Isotopes in Precipitation and the Global Network of Isotopes in Rivers were never designed to systematically represent karst systems. High analysis costs and limited data accessibility have hindered comparative and large-scale studies. WoKaS-Iso addresses this gap by aggregating data from peer-reviewed literature, theses, reports, online archives, and direct contributions from a worldwide network of collaborators, all standardised to a common format.

The compilation builds on the earlier WoKaS database, the first global repository of karst spring discharge records covering more than 400 springs, and extends it with isotope data for 241 of them, 29 of which can be linked directly to their discharge records. The cave component was developed in collaboration with the SISAL working group on speleothem research, drawing primarily on the SISAL_mon_v1 cave monitoring database, which contributed 66 of the 74 caves, and adding 365 individual drip sites. Of the 1007 time series, 389 come from springs and 618 from cave drips. Most records are rich in detail: 79 percent of spring isotope series and 68 percent of cave drip series have resolutions finer than monthly intervals, and 148 springs and 64 caves contain both oxygen-18 and deuterium records.

Assembling the database required considerable detective work. Ninety percent of the spring isotope data and 96 percent of the cave drip isotope data are raw field observations, but the remainder were digitised from figures in published papers using the open-source tool WebPlotDigitizer, with each extracted record validated by replotting and compared against the original graphics. Digitised records are explicitly flagged in the metadata so users can account for the additional uncertainty. The team also paired the isotope records with ancillary environmental variables, including spring discharge, drip rate, precipitation, and rainwater isotopes, and supplemented sites lacking local rainfall data with records from nearby stations of the Global Network of Isotopes in Precipitation, retained only within 25 kilometres of a given spring or cave.

Where observations were unavailable, the database provides modelled inputs from six global products, including the MSWEP precipitation ensemble, the GLEAM evaporation dataset, the ERA5 atmospheric reanalysis, and three precipitation isotope models: IsoGSM, Isoscape, and a sinusoidal seasonal model. The team evaluated these isotope models against site-specific observations at dozens of cave, spring, and meteorological locations. IsoGSM, a general circulation model that incorporates isotopic fractionation and is nudged with reanalysis data, emerged as the most consistent performer, achieving a median root-mean-square error of about 2 per mille for oxygen-18 and coefficients of determination reaching 0.85 or higher at several sites. The sinusoidal model performed markedly worse, with median errors exceeding 3 per mille for oxygen-18 and 29 per mille for deuterium, leading the authors to recommend IsoGSM as the preferred substitute where no local observations exist.

Every site in the database was also assigned a quality score reflecting data availability and source reliability, from raw collaborator-contributed observations and local measurements at the top to digitised records and global model data at the bottom. Half of the cave sites achieved the highest quality class, compared with 21 percent of springs, reflecting the more complete monitoring systems typically installed in caves. The authors note that the quality classification guides appropriate use: the best datasets are recommended for calibrating and validating process-based hydrological models and for paleoclimate reconstruction, mid-tier datasets suit regional comparative studies and sensitivity testing, and lower-tier records help identify data-scarce regions where new monitoring is most needed.

The implications extend well beyond academic hydrology. Karst aquifers are acutely vulnerable to contamination because pollutants can travel rapidly through conduits with little natural filtration, and some European karst regions appear especially sensitive to global warming. By linking isotope-based transit time estimates with hydrological models, researchers can better predict contaminant transport and aquifer vulnerability, informing sustainable water management for the hundreds of millions of people who depend on these systems. The database is publicly archived and machine-readable, with MATLAB extraction scripts and planned web-based visualisations, and the team hopes it will foster collaborative research and encourage data sharing, particularly from underrepresented regions such as Africa, South America, and Southeast Asia. In an era of intensifying drought and growing pressure on groundwater, giving scientists a global, unified view of how water moves through the world’s fastest-flowing aquifers may prove one of the most consequential datasets of the decade.

Subject of Research: A global database of oxygen-18 and deuterium isotope time series in karst springs and cave drip water for studying karst hydrogeology

Article Title: The WoKaS-Iso database: workflow for a global compilation of oxygen-18 and deuterium records in karst springs and cave drip water for enhanced understanding of karst systems

Article References: Zang, Y., Treble, P. C., Yoshimura, K., Pinza, J. G., Zhang, F., Özdemir Çallı, K., Mei, X., Husic, A., Gessert, A., Stroj, A., Andreo, B., Ladouche, B., Stumpp, C., Mance, D., Zagana, E., Huang, F., Sappa, G., Kunstmann, H., Brielmann, H., … Hartmann, A. (2026). The WoKaS-Iso database: workflow for a global compilation of oxygen-18 and deuterium records in karst springs and cave drip water for enhanced understanding of karst systems. Earth System Science Data, 18(10), 7199-7225. https://doi.org/10.5194/essd-18-7199-2026

Image Credits: AI Generated

DOI: 10.5194/essd-18-7199-2026

Keywords: karst hydrogeology, stable isotopes, oxygen-18, deuterium, cave drip water, groundwater, WoKaS-Iso, speleothems, aquifer recharge, water resources, IsoGSM, database