When a river becomes your drinking water
When you open the tap in Tarnów, a charming city in southern Poland with a history dating back to the Middle Ages, chances are that most of the water you are about to drink once flowed in the nearby Dunajec River. Yet between the river and your glass lies an invisible world: a network of sand and gravel that filters, delays, and transforms the river’s water before it reaches the wells.
This natural process, known as riverbank filtration (or, occasionally, induced bank filtration), is a form of Managed Aquifer Recharge (MAR). Across Europe and beyond, it helps cities store and purify water within river-adjacent aquifers. But you may ask, how efficiently do these systems work? How long does it take for river water to reach the wells? Moreover, how can we detect changes before they affect the quality of water that we want to drink?
An operating well field can also be a laboratory
To answer these questions, in cooperation with Tarnów Waterworks Ltd., we “converted” the operating Kępa Bogumiłowicka well field (our riverbank filtration site) near Tarnów into a year-long natural laboratory. Eleven production wells abstract water from a shallow alluvial aquifer beside the Dunajec, providing about one-third of Tarnów’s drinking water.
We monitored the site continuously for twelve months, measuring river and groundwater levels and analysing numerous samples for natural physicochemical fingerprints in water, including stable isotopes of water and chloride concentration, temperature, and electrical conductivity. These environmental tracers enable us to distinguish between river water and native groundwater – each source carrying its own unique chemical fingerprint.
One of the analysed wells (left) and the Dunajec River (right)
Map of the study area along the Dunajec River. Blue arrows indicate the direction of water flow from the river and the land side toward the well field
How we listened to what the water was saying
Every time the Dunajec’s level rose or fell, our sensors recorded a swift response in the groundwater, indicating a strong hydraulic connection between the surface and subsurface. By analysing water isotope ratios and temperature signals, we could even estimate how long it takes for a drop of river water to travel underground to selected wells near the riverbank. The answer surprised us: less than one month (approximately 18 to 22 days). That short journey means that the system is both efficient and sensitive: efficient because water flows quickly through highly permeable sands and gravels, but sensitive because any upstream contamination could also move swiftly toward the wells.
A multi-tracer approach that fits real-world needs
To understand the complex interplay between river water and groundwater, we employed a recently developed tool called the Ensemble End-Member Mixing Analysis (EEMMA). Rather than relying on single measurements, EEMMA uses a time series of a tracer and estimates how much each water source contributes to the mixture. In our study, we applied EEMMA to multiple tracers (stable isotopes of water, Cl⁻, and conductivity) and synthesised the results.
Think of it as mixing colours: by comparing the “shades” of river water and local groundwater, we can tell how much of each pigment appears in each well. This approach proved especially useful when tracer signals overlapped or changed with the seasons — situations common in the real world. Our results showed that river water recharge accounted for over 90% of the yield in wells located near the riverbank (approx. 90–125 m from the river) and around 50% in wells farther inland (approx. 320–390 m from the river). This spatial pattern enables managers to identify which wells are most vulnerable to changes in the river.
What the isotopes revealed about the seasons
Isotopic data indicated that during spring snowmelt, the Dunajec carried water that had originated high in the Tatra Mountains: water depleted in heavy isotopes and easily distinguishable from regional rainfall (or young, regional Tarnów groundwater). That meltwater signature appeared in the wells shortly afterwards, confirming that the upstream river’s mountain sources directly influence Tarnów’s groundwater supply downstream.
By summer, as rainfall dominated the recharge, the isotopic signature shifted toward more enriched (or more positive) delta values, indicating the growing role of local (baseflow) recharge. Such seasonal fingerprints help operators anticipate changes in water quality and design monitoring that captures both short-term events and annual cycles.
Schematic cross-section of the riverbank filtration system. Blue arrows show river water infiltrating toward production wells
From science to practice
Beyond the scientific curiosity, our study also offers practical recommendations, a recipe for anyone managing a riverbank filtration site, for example:
We also provide the dataset and R scripts openly on the Zenodo repository, allowing others to test, reuse, or adapt the workflow to their own sites.
Why do these findings matter?
Across Europe, riverbank filtration supplies drinking water to millions of people — from the Rhine to the Danube. Yet many sites still rely on sparse monitoring required only by regulation. Our results show how relatively low-cost sensors and open-source analytics can turn such systems into early-warning networks that protect both water quantity and quality.
Understanding these hidden exchanges between rivers and aquifers is also crucial for effective climate adaptation and mitigation. As floods and droughts become more frequent, cities will need flexible water systems that store water underground when it’s abundant and recover it when it’s scarce. MAR, when guided by smart monitoring, offers exactly that resilience.
Lessons learned along the way
One of the surprises during fieldwork was such a rapid response of the aquifer to changes in the river level or system operational regime. During a short maintenance shutdown, for instance, we observed a sudden rise in conductivity within an hour—a clear signal that more native groundwater had entered the wells. Moments like this remind us that aquifers, in a sense, are “alive” organisms, especially the shallow, highly permeable ones; they respond and adapt on timescales that we can now observe in real time. Working closely with Tarnów Waterworks Ltd. also underscored the importance of collaboration between scientists and public utility operators. Access to real-world sites and operational data makes research directly applicable to daily water-supply decisions.
The bigger picture
Our findings go beyond a single well field. They demonstrate that by integrating continuous monitoring, environmental tracers, and open-data tools, we can “see” the hidden flow paths that sustain our drinking water resources. Ultimately, riverbank filtration demonstrates that the boundary between surface and groundwater is not a fixed line, but a dynamic interface—one that filters, connects, and reminds us of the profound interconnection of the Earth’s systems.
And finally, it is fitting to quote the motto of Tarnów Waterworks: “Nothing without good water”
During the field work …
Further reading
Janik, K., Rein, A., and Sitek, S.: Towards efficient management of riverbank filtration sites: new insights on river–groundwater interactions from environmental tracers and high-resolution monitoring, Hydrology and Earth System Sciences, 29, 5893–5911, https://doi.org/10.5194/hess-29-5893-2025, 2025.
Edited by Bettina Schaefli;
A note from the editorial team: This is the first guest blogger contribution received following a recent innovation in the Copernicus Publishing System: upon acceptance of your paper in one of our journals, authors are invited to consider if they would like to turn their science paper into a blog post. The present contribution nicely illustrates how some research papers can be effectively turned into a blog post for a wider audience.
