Wastewater-aware baseflow separation sharpens pollutant load estimates in a Hungarian catchment

Estimating how much pollution a river carries is harder than it sounds, and in a small Hungarian catchment it is harder still, because more than 90 percent of the water flowing in the upper reach of the Koppány stream is treated wastewater rather than natural runoff. A new study published in Applied Water Science presents a modified baseflow separation framework designed for exactly this kind of system, where a continuous point-source discharge fundamentally alters the flow regime. The work, led by Tímea Lajkó with Adrienne Clement and Máté Krisztián Kardos, was applied to the 661 square kilometre Koppány catchment in Somogy County, West Hungary, a landscape dominated by arable agriculture and shaped by a single, unusually large wastewater input near the stream’s headwaters.

The core of the study is a comparison of three established digital filter methods for separating total streamflow into a slowly varying groundwater-derived baseflow component and a rapidly responding surface runoff component: the Lyne–Hollick filter, the Eckhardt filter, and the Rimmer–Hartmann method, which adds a geochemical tracer to constrain the calibration. The researchers used electrical conductivity, recorded at five-minute intervals by online sensors at two gauging stations, Törökkoppány upstream and Tamási downstream, as the tracer. Where the standard Rimmer–Hartmann approach failed, the team modified it by introducing wastewater as an explicit third end-member in the conductivity-based mixing equation, alongside natural baseflow and precipitation-driven high flow.

The failure of the unmodified method is instructive. When applied to the Koppány data, the standard Rimmer–Hartmann procedure indicated that baseflow constituted 100 percent of streamflow at both stations, a physically unrealistic result that contradicted prior experience in the catchment. The problem traced back to the beta parameter, which represents the maximum attainable baseflow index in the underlying Eckhardt filter: the optimization produced excessively high beta values, overestimating the baseflow contribution. The culprit was the wastewater itself. Roughly 5,200 cubic metres of treated effluent from the Balatonlelle wastewater treatment plant, a facility with a design capacity of 104,000 population equivalents, is discharged daily into the upper Koppány, and because the effluent carries a distinct electrical conductivity signature, it distorted the two-component mixing calculation that assumes only groundwater and surface runoff are present.

The fix was conceptually simple but consequential. The researchers added the wastewater flow rate as a third term in the baseflow equation and its characteristic conductivity as a third end-member in the mixing equation, treating wastewater discharge and conductivity as constant operational parameters during calibration. With this three-component formulation, the modified method yielded physically realistic baseflow index values of 0.80 at Törökkoppány and 0.83 at Tamási. At Törökkoppány, the calibrated beta of 0.80 coincided with the value recommended by the Eckhardt method, so the two approaches produced identical estimates there, and the Lyne–Hollick filter, run with the widely used default alpha of 0.925, gave 0.82 at Törökkoppány and 0.92 at Tamási. The Eckhardt filter returned 0.80 at both stations, a convergence the authors interpret as corroboration of the modified result.

Separating flow is only half the task; the separation must then be translated into a sampling and load estimation strategy. The team classified each time unit as an event or non-event period by comparing the calculated baseflow share against a threshold, and selected those thresholds through a sensitivity analysis across nine candidate values per station, ranging from 0.70 to 0.90. Because the true baseflow index of the catchment is unknown, the long-term regional reference of 0.72 to 0.75 reported by Jolánkai served as the benchmark. A threshold of 0.77 was selected for Törökkoppány, yielding a baseflow index of 75 percent with 287 non-event days, and 0.82 was selected for Tamási, yielding 76 percent with 279 non-event days. The analysis also revealed how differently the two stations behave: at Törökkoppány, raising the threshold from 0.70 to 0.90 shifted classification from 342 non-event days to 326 event days, while at Tamási the transition was far more abrupt, from 363 non-event days at 0.70 to just 13 at 0.90.

The load estimation itself rested on a hybrid monitoring programme combining weekly grab samples under low-to-medium flow conditions, composited into six laboratory analyses per station over the year, with event-based automatic sampling triggered by rising stage, which captured 25 to 30 samples per station during high-flow events. Continuous monitoring of water level, temperature, and electrical conductivity ran in parallel, with discharge derived from a rating curve. Annual loads were then calculated by multiplying the mean low-to-medium flow concentration by the total non-event water volume and the mean event concentration by the total event water volume, summed across the investigated period. The pollutant groups covered nutrients, metals, pharmaceuticals, and the per- and polyfluoroalkyl substances PFOS and PFOA.

The load results reveal sharply contrasting transport pathways. At Törökkoppány, most metals were transported predominantly during events, with total annual specific loads of 1.80 kilograms per square kilometre for nickel, 2.22 for chromium, and 1.17 for zinc, and particulate metal fractions showed 71 to 92 percent of annual loads moving during events, consistent with erosion-driven transport from the arable hillslopes that cover 79 percent of the catchment. Dissolved fractions behaved differently: 82 percent of the dissolved arsenic load and 95 percent of the dissolved chromium load were attributed to non-event conditions, a pattern the authors suggest points to a geogenic origin linked to local geology. Concentration data reinforced the picture, with total nickel at Törökkoppány rising from 19.7 micrograms per litre under low-to-medium flow to 137 micrograms per litre during high flow, driven largely by the particulate phase.

Organic micropollutants told a different story. Pharmaceutical residues behaved as expected for a continuous wastewater source: 89 percent of the annual carbamazepine load and 93 percent of the diclofenac load at Törökkoppány, and 89 percent and 88 percent respectively at Tamási, were transported under non-event conditions. PFOS was strongly non-event-dominant at both stations, with 70 percent of its annual load at Törökkoppány and 81 percent at Tamási attributed to baseflow. PFOA was more ambiguous, distributed roughly evenly between states at Törökkoppány, where its concentrations increased during events, contrary to the dilution expected for a wastewater-derived compound, prompting the authors to suggest contributions from atmospheric deposition, shallow groundwater, or mobilisation from soil and suspended particles. For nutrients, dissolved orthophosphate was predominantly non-event-associated at both stations, 84 percent of the annual load at Törökkoppány and 80 percent at Tamási, reflecting sustained wastewater contributions to low flow, while particulate phosphorus shifted from near-even splitting at the upstream station to 56 percent non-event transport at Tamási, where upstream fishponds attenuate flood waves and retain sediment.

The overall downstream contrast was substantial: event conditions accounted for over 74 percent of the total micropollutant load at Törökkoppány but approximately 61 percent at Tamási, a shift the authors attribute to the flood-attenuating effect of the fishponds, which produce lower peak discharges but prolonged flood durations at the downstream station. The authors are explicit about the method’s limitations. Load estimates carry inherent uncertainty from the binary classification of event and non-event periods, which imposes a discrete threshold on a continuous hydrological transition, and from the assumption of constant wastewater conductivity and discharge during calibration, even though effluent volume fluctuates strongly with the seasonal holiday population of the Lake Balaton region. Direct independent validation through groundwater level records, stable isotopes, or chloride tracers was unavailable for the catchment, so the reliability argument rests instead on convergence with independent digital filters and on regional hydrogeological evidence, including isotope studies confirming upwelling of older, deeper groundwater near the valley bottoms.

What the framework offers, the authors argue, is a practical and transferable tool for managed catchments where a single identifiable point source with a distinct conductivity signature dominates the flow regime. It relies solely on electrical conductivity, a low-cost, continuously monitored parameter, and operational wastewater discharge records, avoiding the logistical demands of isotopic or chloride-based campaigns. The approach is most applicable where the conductivity contrast between effluent, precipitation, and natural groundwater is large; where multiple point sources exist or effluent conductivity approaches that of natural baseflow, additional end-members or alternative tracers would be required. For the many river basins where treated sewage constitutes a large and persistent share of streamflow, the study suggests that ignoring that wastewater component in baseflow separation risks systematically misattributing where pollutants come from and how they travel.

Subject of Research: A modified baseflow separation and stratified sampling framework for estimating pollutant loads in a wastewater-impacted Hungarian catchment

Article Title: Baseflow separation combined with stratified sampling serving the accurate estimation of pollutant loads—the example of a small catchment with high anthropogenic pressures

Article References: Lajkó, T., Clement, A., & Kardos, M. K. (2026). Baseflow separation combined with stratified sampling serving the accurate estimation of pollutant loads—the example of a small catchment with high anthropogenic pressures. Applied Water Science, 16(11), Article 293. https://doi.org/10.1007/s13201-026-02986-9

Image Credits: AI Generated

DOI: 10.1007/s13201-026-02986-9

Keywords: baseflow separation, pollutant load estimation, Koppány stream, electrical conductivity, Rimmer-Hartmann method, stratified sampling, wastewater discharge, baseflow index, nutrients, PFAS, pharmaceuticals, Hungary