For decades, hydrologists have treated rising groundwater levels as a proxy for fresh water entering an aquifer. When winter rains fall and water tables climb, the standard interpretation is that new water has arrived, replenishing the storage that wells and springs draw upon. This logic underpins the water table fluctuation method, one of the most commonly applied recharge estimation techniques. But in layered, faulted aquifer systems, the authors argue, the assumption breaks down. Water levels can also respond to pressure transmission through connected fractures, redistribution of stored water between layers, and mixing of waters of different ages. A rising borehole may be reporting a hydraulic message traveling through the system, not a delivery of new water.

To test this idea, the researchers focused on the G50B sub-catchment, locally known as the Nuwejaars River catchment, a roughly 760 square kilometer mountainous headwater zone within the larger Heuningnes Catchment. The area lies in the southern branch of the Cape Fold Belt, a geological province of folded, faulted, and fractured sedimentary rocks that includes the Table Mountain Group, the Bokkeveld Group, and younger Bredasdorp deposits. These units form a heterogeneous multilayer aquifer in which fractured secondary aquifers and unconsolidated primary aquifers coexist with sharply contrasting hydraulic properties. A Mediterranean climate concentrates 65 to 75 percent of annual rainfall in the winter months of May through October, providing a strong seasonal signal against which aquifer responses can be measured.

The team’s central innovation was to divide the catchment into four hydrogeological process domains, defined independently of the datasets used in the analysis. Recharge-dominated systems experience active infiltration and short circulation pathways. Structurally controlled domains are shaped by fault- and fracture-mediated groundwater movement. Residence-time-dominated systems are characterized by prolonged storage and extended flow paths. Transitional domains, as the name suggests, host interacting processes. Because this classification was developed before the groundwater-level and age data were examined, it provided an a priori framework that reduced the risk of circular reasoning in the interpretation.

The age of the groundwater itself was constrained using tritium, a radioactive isotope of hydrogen whose concentrations in the atmosphere spiked dramatically during the nuclear bomb tests of the 1960s and have since declined. The researchers reconstructed a complete annual tritium input function for the region from 1946 to 2025, drawing on pre-bomb ice-core baselines of roughly 4.6 tritium units, records from the International Atomic Energy Agency’s Global Network of Isotopes in Precipitation, and regional Western Cape measurements. Groundwater samples, enriched by electrolysis and counted by liquid scintillation at iThemba Laboratories in Johannesburg, were then compared against this atmospheric record to estimate how long the water had been underground.

Crucially, the team applied two different age models. Under simple piston-flow assumptions, which treat water as traveling along a single path without mixing, apparent ages ranged from essentially modern to at least 60 years. Under the more realistic exponential-piston flow model, which explicitly incorporates mixing between young recharge and older stored water, mean transit times stretched from as little as 2 years to as much as 80 to 100 years. The gap between the two approaches was sometimes enormous. In several boreholes, including BH13, BH15, BH16, BH17, BH21, BH23, BH24, and BH27, the discrepancy exceeded 30 years, and in some locations it surpassed 45 years. The ratio of piston-flow age to modeled transit time ranged from about 0.36 to 1.30, demonstrating that simplified age estimates can either understate or overstate true residence times depending on local mixing conditions.

The most striking results emerged when groundwater ages were paired with hourly water-level measurements collected by data loggers between December 2020 and November 2021. Response rates across the monitoring network ranged from 0.022 to 0.095 meters per day, and fluctuation amplitudes spanned 1.02 to 4.27 meters. Yet when the researchers plotted response rate against mean transit time, the relationship collapsed. The correlation was weak and statistically non-significant, with a Pearson’s r of minus 0.307 and a p-value of 0.460. Borehole BH22 recorded the fastest response in the network, 0.095 meters per day, despite a mean transit time of only 3 years, while BH23 responded more slowly at 0.049 meters per day even though its water carried a transit time of roughly 60 years. In the transitional domain, BH13 and BH14 showed nearly identical response rates of 0.054 and 0.056 meters per day, but their mean transit times differed fourfold, at 60 and 15 years respectively.

Each domain told its own story. Recharge-dominated systems displayed strong seasonal water-level responses, yet even here older components persisted; BH23, in the mountainous upper catchment, showed a fluctuation exceeding 3 meters while harboring water with a 60-year transit time. Structurally controlled domains provided the clearest evidence of decoupling, with rapid hydraulic responses coexisting with residence times of 80 to 100 years, a pattern the authors attribute to pressure signals propagating through fractures while actual water transport remains slowed by fracture-matrix interactions. Residence-time-dominated systems behaved as expected, with old water and subdued, attenuated hydrographs. The transitional domain exhibited the widest spread of ages, from 2 to 80 to 100 years, reflecting the coexistence of multiple recharge pathways and nested flow systems.

The practical implications are considerable. Previous research has shown that water table fluctuation methods can overestimate recharge by roughly 20 to 34 percent where delayed recharge and vadose-zone processes complicate water-level behavior, and the Heuningnes results suggest that structural connectivity can push such errors further. In a catchment where fault-controlled springs and artesian conditions have been documented, a rising water table may simply reflect a well-connected aquifer transmitting pressure, not a reservoir being actively refilled. For a country facing increasing water stress, misreading that signal could mean overestimating the sustainable yield of an aquifer and allocating more water than nature replaces.

The authors are careful to acknowledge the limitations of their approach. Mean transit times were derived from a single tracer with a fixed mixing parameter, and gaps in the historical tritium record required constrained interpolation, introducing uncertainty that is greatest for intermediate-age groundwater. Substantial variability also occurred within individual domains, confirming that local geology modifies broader system-scale patterns. Future work, they suggest, should combine tritium with radiocarbon, noble gases, and chlorofluorocarbons to better constrain the full spectrum of transit times. Even so, the core message stands: in structurally complex multilayer aquifers, groundwater age is best understood as a distribution rather than a single number, and a responsive borehole is not necessarily a replenished one. Distinguishing the speed of a hydraulic signal from the speed of the water itself may prove essential for groundwater management wherever fractured rock and layered aquifers dominate.

Subject of Research: Hydraulic-age decoupling and groundwater recharge assessment in the structurally complex multilayer aquifer system of the Heuningnes Catchment, South Africa

Article Title: Old water, rapid responses: Process-based hydrogeological domains reveal hydraulic–age decoupling and implications for recharge assessment in a complex aquifer system of the Heuningnes Catchment, South Africa

Article References: Mqhayi, S., & Kanyerere, T. (2026). Old water, rapid responses: Process-based hydrogeological domains reveal hydraulic–age decoupling and implications for recharge assessment in a complex aquifer system of the Heuningnes Catchment, South Africa. Hydrogeology Journal. https://doi.org/10.1007/s10040-026-03178-0

Image Credits: AI Generated

DOI: 10.1007/s10040-026-03178-0

Keywords: groundwater recharge, tritium dating, mean transit time, hydrogeology, Cape Fold Belt, aquifer heterogeneity, water table fluctuation, piston flow model, exponential-piston flow model, South Africa, groundwater age, hydraulic connectivity