Deliverable D2.4 - Guidance document on fate, transport and exposure for PMT's in the environment

Project ID N°:101036449 Call: H2020-LC-GD-2020-3 Topic: LC-GD-8-1-2020 - Innovative, systemic zero-pollution solutions to protect health, environment, and natural resources from persistent and mobile chemicals Preventing Recalcitrant Organic Mobile Industrial chemicalS for Circular Economy in the soil-sediment-water System Start date of the project: 1st November 2021 Duration: 42 months D2.4 – Guidance document on fate, transport and exposure for PMT’s in the environment Main authors: Zessner Matthias, Baldwin Dwight, del Val Alonso Laura, Derx Julia, Devau Nicolas, Janssen Gijs, Jou Claus Sònia, Kittlaus Steffen, Knoche Franziska, Liu Meiqi, Markus Arjen, Valstar Johan, Meesters Joris, Meijers Erwin, Obeid Ali A.A., Oudega Thomas James, Pathak Devanshi, Sprenger Christoph, van Gils Jos, Wicke Daniel, Wintersen Arjen, Zhiteneva Veronika, Groot Hans Lead Beneficiary: TU Wien Type of delivery: R Dissemination Level: PU Filename and version: PROMISCES_D2.4_Guidance-document (Version 1.0) Website: https://promisces.eu/Results.html Due date: 31/10/2024 Authors and their affiliations: Zessner Matthias 1 , Baldwin Dwight 7 , del Val Alonso Laura 6 , Derx Julia 2 , Devau Nicolas 3 , Janssen Gijs 4 , Jou Claus Sònia 6 , Kittlaus Steffen 1 , Knoche Franziska 7 , Liu Meiqi 1 , Markus Arjen 4 , Valstar Johan 4 , Meesters Joris 5 , Meijers Erwin 4 , Obeid Ali A.A. 2 , Oudega Thomas James 2 , Pathak Devanshi 4 , Sprenger Christoph 7 , van Gils Jos 4 , Wicke Daniel 7 , Wintersen Arjen5, Zhiteneva Veronika7, and Groot Hans4 1 TU Wien, Institute of Water Quality and Resource Management, Karlsplatz 13, Vienna, Austria 2 TU Wien, Institute of Hydraulic Engineering and Water Resources Management, Karlsplatz 13, Vienna, Austria 3BRGM, French Geological Survey, 2 avenue Claude Guillemin, Orléans France 4Deltares. Boussinesqweg 1, Delft, the Netherlands 5 National Institute for Public Health and the Environment (RIVM), Bilthoven, the Netherlands 6Eurecat, Plaça de la Ciència 2, Manresa, Catalonia, Spain 7Berlin Centre of Competence for Water (KWB), Grunewaldstr. 61-62, Berlin, Germany Authors of annexes are named at the beginning of the annexes. ©European Union, 2025 This work is licensed under Creative Commons Attribution 4.0 International. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/ No third-party textual or artistic material included on the publication without the copyright holder’s prior consent to further dissemination by other third parties. Reproduction is authorized provided the source is acknowledged. Disclaimer The information and views set out in this report are those of the author(s) and do not necessarily reflect the official opinion of the European Union. Neither the European Union institutions and bodies nor any person acting on their behalf may be held responsible for the use which may be made of the information contained therein. Document History This document has been through the following revisions: Version Date Author/Reviewer Description 0.1 13.01.2025 Matthias Zessner et al. First draft for review 0.2 31.01.2025 Matthias Zessner et al. Second draft for validation 0.3 06.02.2025 Steffen Kittlaus et al. Third draft for quality control 0.4 14.02.2025 Steffen Kittlaus et al. Fourth draft for approval 1.0 25.02.2025 Matthias Zessner Final Version for distribution Authorisation Authorisation Name Status Date Review Dominique Guyonnet Reviewer 20.01.2025 Validation Hans Groot WP2 leader 05.02.2025 Quality Control Floriane Sermondadaz IEIC Manager 10.02.2025 Approval Julie Lions Project Coordinator 27.02.2025 Distribution This document has been distributed to: Name Title Version issued Date of issue TU Wien WP2 partner Version 1.0 03.03.2025 BRGM WP2 partner Version 1.0 03.03.2025 Deltares WP2 partner Version 1.0 03.03.2025 RIVM WP2 partner Version 1.0 03.03.2025 Eurecat WP2 partner Version 1.0 03.03.2025 KWB WP2 partner Version 1.0 03.03.2025 D2.4 – Guidance document on fate, transport and exposure for PMT’s 3 Executive Summary Models are used in exposure assessment for a number of reasons. They can help map the temporal and spatial variability of exposure, exposure pathways and exposure routes, and support risk assessment for water bodies where monitoring is lacking. They can be used to identify sources and pathways responsible for current exposures and to assess the impact of potential future developments of persistent, mobile, and toxic chemicals (PMT) exposures in surface water and groundwater. Such scenario assessment may include changes in PMT use, effects of pollution control measures, accidental spills or climate change. The scope of this document, produced as part of the H2020 PROMISCES project, is to provide guidance for applications of models with a specific focus on model trains for the assessment of exposure to PMTs as part of the predictive risk assessment related to surface and groundwater. This document explains the basic concepts of specific models and how best to use them in model trains in the framework of a tiered approach. The intention is to inform users and interested stakeholders about what needs to be considered when using different methods, what is the best use of specific models, what are the best combinations in model trains and what are their current limitations. The guidance document presents (i) “screening level” models for the assessment of regional exposure of groundwater from soil pollution and for the assessment of general exposure of air, soil and water at local, regional or global scales, (ii) spatial and temporal explicit approaches for the identification of pollution plumes in the soil-groundwater continuum and (iii) model train applications for the catchment – river – river bank filtration – drinking water continuum. Exposure of surface water and groundwater to PMT depends on the use patterns and the environmental fate of the chemicals. Emission, fate and transport models incorporate driving factors into documented algorithms. The extent to which a substance persists in surface water can, for instance, be calculated with the “SimpleBox - Aquatic Persistence Dashboard”, based on its physical-chemical characteristics. The presented approach for deriving generic risk limits for soils shows that, depending on regional variations in geo(hydro)logical conditions, the high mobility of some PFAS could lead to strict requirements for materials applied on soil. For the soil-groundwater continuum, a novel model train is presented which accounts for the main physical and chemical processes controlling the fate and transport of PFAS. For sorption and degradation reactions, several formalisms can be used, allowing one to select the most appropriate according to the PFAS molecular properties and the characteristics of the simulated domain. The results issued from these modelling applications indicate the key role of correctly identifying the main physical, chemical and biological processes controlling fate and transport of PFAS in the studied domain to build a robust conceptual model. To increase the robustness of the model, a thorough model calibration must be performed, preferably using time series measurements of the PFAS concentration in the pore solution at different locations of the contaminated site. D2.4 – Guidance document on fate, transport and exposure for PMT’s 4 The results confirm the key role of the unsaturated zone in the transfer and long-term migration of PFAS. Nonlinearity and nonideality of sorption reactions were expected for a broad range of PFAS, suggesting using more complex numerical formalism than linear isotherms. Considering the key role of capillary fringe displacement on PFAS transport in the unsaturated zone, the model train seems to be very efficient in performing PFAS simulations, as it can explicitly describe water flow and solute transport at the interface between the unsaturated and saturated zones, avoiding the main pitfall encountered in other numerical approaches. The combination of stand-alone models in model trains expands the scope that can be covered in the context of a catchment – river – riverbank filtration – drinking water continuum for exposure assessment of surface waters and bank filtered drinking water. Model trains can combine individual models either in a complementary way or in a sequence. A complementary combination may either compare models of different complexity to find out which level of complexity (and associated effort) is needed to answer which questions, or may compare different models with their different strengths and weaknesses in parallel to assess uncertainties and/or use models for scenario evaluation according to their specific capabilities. A sequential combination facilitates a broader application in terms of content and at different spatial resolutions. Clearly defined interfaces are essential for a successful implementation. Examples of model trains for selected PFAS are presented for the catchment-river interaction in the urban context of the Berlin case and for the whole catchment – river – riverbank filtration – drinking water continuum on the scale of the Upper Danube Basin. The Berlin case demonstrates the application of the sequential model train by combining a city emission model with a city surface water fate and transport model to assess the resulting exposure to PFAS in the city surface waters. The Danube case demonstrates the application of a sequential model train for exposure assessment of bank filtered drinking water by combining large-scale catchment-scale emission models with different types of bank filtration fate and transport models for specific locations in the catchment. In addition, it also demonstrates complementary application by comparing emission models with different strengths and weaknesses for the assessment of multiple scenarios on the catchment scale and different levels of complexity for the fate and transport modelling of bank filtration. The model train has been successfully applied for 10 different PFAS-substances including the assessment of a large range of scenarios. Current limitations for exposure assessment of PFAS at river basin scale require improvement in scientific understanding as well as additional efforts in administrative data collection and inventory development. Current results of the exposure assessment show the very high relevance of legacy pollution from use of fire-fighting foams or from old municipal landfills. On the administrative level, there is a strong need for improved identification and harmonized inventorying of contaminated sites at national and international (EU) level. The lack of robust, openly available information on production, import-export and therefore use volumes of PFAS at national and EU level is strongly hampering exposure assessment. A major effort is urgently needed to provide this information, as it is decisive for a sound environmental exposure assessment, not only for surface water and groundwater. In regard to scientific advances, there is a need for more and better understanding of the extent of local groundwater pollution, particularly due to the application of fire-fighting foams or to the presence of municipal landfills. Further improvement of the scientific knowledge about the fate of PFAS in the environment, including their partitioning between different phases (air, water, solids) and the transformation of the so called “precursors” into stable “end-products” D2.4 – Guidance document on fate, transport and exposure for PMT’s 5 like PFOA, PFOS and short-chain substances is needed to enlarge the number of PFAS that can be included into the exposure assessment. A reproducible and standardised analytical parameter for “total PFAS” or even “total toxicity of PFAS” would be needed to address all relevant PFAS in a combined way as it is a focus of Workpackage 1 of the H2020 PROMISCES project (Togola et al. 2024; Behnisch et al. 2024). D2.4 – Guidance document on fate, transport and exposure for PMT’s 6 Contents 1. Introduction 19 1.1. Emission, fate and transport modelling for exposure assessment . . . . . . . . . 19 1.2. Scope of the guidance document . . . . . . . . . . . . . . . . . . . . . . . . . . 21 1.3. Basics of exposure assessment and environmental fate . . . . . . . . . . . . . . 24 2. Screening level models 27 2.1. Introduction...................................... 27 2.2. The SimpleBox - Aquatic Persistence dashboard . . . . . . . . . . . . . . . . . 28 2.3. Deriving generic risk limits for soils . . . . . . . . . . . . . . . . . . . . . . . . . 34 3. Soil - Groundwater interaction 40 3.1. Generalaspects.................................... 40 3.2. Model train combining Hydrus and MODFLOW/MT3DMS . . . . . . . . . . . 42 3.3. General explanation of case studies . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.4. Application for the PluriMetric Pilot experiment . . . . . . . . . . . . . . . . . 48 3.5. Application of the model train in the CS#7 AFFF-polluted aquifer . . . . . . . 58 3.6. Limitations and further outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4. Catchment – river – riverbank filtration – drinking water 64 4.1. Generalaspects.................................... 65 4.2. Emission models on catchment scale in the context of model trains . . . . . . . 66 4.3. Emission model in an urban context . . . . . . . . . . . . . . . . . . . . . . . . 71 4.4. Fate and transport modelling at bank filtration sites . . . . . . . . . . . . . . . 73 4.5. Model train application in an urban setting of Berlin . . . . . . . . . . . . . . . 79 4.6. Model train application in a large catchment . . . . . . . . . . . . . . . . . . . 84 4.7. Limitations and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 5. Discussion and conclusions 111 5.1. Background and scope of the guidance document . . . . . . . . . . . . . . . . . 111 5.2. Pollution plumes in the soil-groundwater continuum . . . . . . . . . . . . . . . 112 5.3. Modelling of the catchment – river – riverbank filtration – drinking water continuum 112 Bibliography 115 Annexes 120 A. Substance properties to be used in the SB-AP Dashboard 121 B. PluriMetric Pilot experiments 125 B.1. Key concepts to simulate PFAS fate and transport in SGW continuum . . . . . 125 D2.4 – Guidance document on fate, transport and exposure for PMT’s 7 Contents B.2. Extended description of the PMP experiment . . . . . . . . . . . . . . . . . . . 127 B.3. Hydraulic, solute transport and reactive properties for simulations of the CS#6 131 B.4. Simulations of the spatial and temporal pattern of the non-reactive tracer concentration during the PMP experiment . . . . . . . . . . . . . . . . . . . . . 136 C. Large scale investigations in CS#7 AFFF-polluted aquifer 139 C.1.Conceptualmodel .................................. 139 C.2.Modelset-up ..................................... 140 C.3. Results from the unsaturated zone model . . . . . . . . . . . . . . . . . . . . . 143 C.4. Results from fully saturated model . . . . . . . . . . . . . . . . . . . . . . . . . 144 C.5.Conclusions...................................... 146 D. Catchment Model MoRE for the upper Danube basin 150 D.1.Overview ....................................... 150 D.2.ModelledPFASs ................................... 151 D.3.ModelledPathways.................................. 152 D.4.BasicModelSetup .................................. 153 D.5. Approach for Validating the Model . . . . . . . . . . . . . . . . . . . . . . . . . 154 D.6.ModelValidation................................... 158 D.7. Modelled Pathway Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 D.8.RiskMaps....................................... 168 D.9.Acknowledgements .................................. 170 E. PROMISCES watershed model for PM substances (PPM) 173 E.1.Introduction...................................... 173 E.2.Methods........................................ 174 E.3.Results......................................... 185 E.4.Acknowledgements .................................. 194 F. CS#2: MODFLOW model of the Szentendre Island 197 G. Scenario modelling for the upper Danube basin 203 G.1. General Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 G.2.Referencesituation.................................. 204 G.3.BaselineforScenarios ................................ 204 G.4. Accidental spill scenario (AC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 G.5. Water pollution control 1 (WPC1) . . . . . . . . . . . . . . . . . . . . . . . . . 209 G.6. Water pollution control 2 (WPC2) . . . . . . . . . . . . . . . . . . . . . . . . . 210 G.7.SummaryofScenarios ................................ 213 G.8.Scenarioresults.................................... 214 D2.4 – Guidance document on fate, transport and exposure for PMT’s 8 List of Figures 1.1. Losses of chemicals to the environment via stages of the products life-cycle . . 24 1.2. Conceptual multimedia box model for environmental fate and exposure . . . . 26 2.1. Aquatic persistence for eight selected substances occurring as dissolved species or the sum of dissolved and sorbed species on a regional, continental and global scale as calculated with the SB-AP Dashboard . . . . . . . . . . . . . . . . . . 31 2.2. Overview of sensitivity plots for substances with NFBS, 6:2 FTOH and 6:2 FTAB for aquatic persistence (s) vs water degradation rate constant (log s -1 ) and log KOW ..................................... 32 2.3. Conceptual model for the groundwater variant . . . . . . . . . . . . . . . . . . 38 2.4. Conceptual model for the surface water variant . . . . . . . . . . . . . . . . . 38 3.1. Sketch of the model train developed to simulate fate and transport of PFAS in soil-groundwater continuum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.2. Measured and simulated concentrations in 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam over time in pore solutions at five depths below the infiltration square during thePMPexperiment................................. 54 3.3. Measured and simulated concentrations in 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam over time in pore solutions at three depths in the saturated zone in two locations downstream to the infiltration square in the main flow path during the PMP experiment. ..................................... 56 3.4. Vertical distribution of main HYDRUS 1D modelled variables in the unsaturated profile......................................... 59 3.5. Downgradient distribution of most abundant PFAS at the site in the saturated groundwater...................................... 60 4.1. Schema of the circular economy route “semi-closed water cycle” . . . . . . . . 66 4.2. Overview of riverine load (Tier 2), pathway-oriented (Tier 3) and source-oriented (Tier4)approaches.................................. 68 4.3. Considered substance sources and input paths of watercourse pollution . . . . 72 4.4. Generic PMT substance attenuation heatmaps and breakthrough curves . . . 77 4.5. Left side: Monthly PFOA loads by path into Berlin’s surface waters, right side: Contributions of the separate sewer system (Sep. sewer), wastewater treatment plants (WWTP) and combined sewer overflows (CSO) to annual rainwater and wastewaterdischarges. ............................... 80 4.6. Layout of the Berlin surface water model . . . . . . . . . . . . . . . . . . . . . 81 4.7. Schematic representation of the Flughafensee as a column model . . . . . . . . 82 4.8. Simulated concentrations of PFOA in Berlin watercourses . . . . . . . . . . . 83 4.9. Map of the Danube catchment upstream of Budapest . . . . . . . . . . . . . . 85 4.10. Map of bank filtration sites at Budapest . . . . . . . . . . . . . . . . . . . . . 86 D2.4 – Guidance document on fate, transport and exposure for PMT’s 9 List of Abbreviations EQS Environmental Quality Standards according to Environmental Quality Standard Directive EQSD Draft of the environmental quality standard directive (EC COM(2022) 540 final) EU European Union fOC Fraction of organic carbon FAO Food and Agriculture Organization of the United Nations FP floodplain attenuation model approach Fract. Dimensionless fraction of adsorption sites with instantaneous sorption FTOH Fluorotelomer alcohol FTSAAm N,N-Dimethyl-3-((perfluorohexyl)ethylsulfonyl)aminopropanamine N-oxide gBF generic Bank Filtration Model GenX Commercial name of HFPODA GLRLs generic risk limits for soils GW Groundwater GWD Draft of the groundwater directive (EC COM(2022) 540 final) HFPO-DA hexafluoropropylene oxidedimer acid / perfluoro-2propoxypropanoic acid I Tortuosity parameter in the conductivity function IMOD-WQ iMOD-Water Quality; graphical User Interface + an accelerated Deltares-version of MODFLOW (see below) Kd Linear adsorption isotherm for soil or sediment KfHydraulic conductivity Koc organic carbon-water partition coefficient Kow octanol/water partitioning coefficient KGE Kling-Gupta model efficiency coefficient (Gupta et al. 2009) Ks Saturated hydraulic conductivity KW kinematic wave model approach L length LOQ Analytical Limit of Quantitation M mass MAE Mean Absolute Error mNSE Modified Nash-Sutcliffe model efficiency coefficient MODFLOW Modular three-dimensional finite-difference ground-water flow model MODPATH a particle-tracking postprocessing package that was developed to compute threedimensional flow paths using output from steady-state or transient ground-water flow simulations by MODFLOW D2.4 – Guidance document on fate, transport and exposure for PMT’s 16 List of Abbreviations MoRE Modeling of Regionalized Emissions MSL Mean Sea Level MT3DMS Modular Three-Dimensional Multispecies Transport Model for Simulation MW Monitoring Well n Parameter n in the soil water retention curve N-EtFOSAA N-ethyl perfluorooctane sulfonamido acetic acid ne effective porosity NFBS 1,1,2,2,3,3,4,4,4-nonafluoroN-methyl -1-Butanesulfonamide nRMSE Normalized root mean square error NSE Nash-Sutcliffe model efficiency coefficient Nu Langmuir coefficient p.e. Population equivalent PFAA polyfluoralkyls acids PFAS Perand polyfluoroalkyl substances PFBA perfluorobutanoic acid PFBS perfluorobutane sulfonic acid PFCA Perfluoroalkyl carboxylic acid PFDA perfluorodecanoic acid PFDS Perfluorodecane sulfonic acid PFHpA perfluoroheptanoic acid PFHxA perfluorohexanoic acid PFHxS perfluorohexane sulfonic acid PFNA perfluorononanoic acid PFNS perfluoronoane sulfonic acid PFOA perfluorooctonoic acid PFOA-eq PFOA toxicity equivalents of PFAS substances according to (EC COM(2022) 540 final) PFOS perfluorooctane sulfonic acid PFOSA perfluorooctanesulfonamide PFPeA perfluoropentanoic acid PFPeS perfluoropentane sulfonic acid PFSA Perfluoroalkane sulfonic acid PHS Priority Hazardous Substances (WFD) PM Persistent and mobile chemicals PMP PluriMetric Pilot PMT Persistent, mobile and toxic chemicals PPM PROMISCES watershed model for PM substances PS Priority Substances (WFD) PSA probabilistic sensitivity analyses PW Pumping Well Q95 Low flow with an exceedance of 95 % of the flows of a period Qr Residual soil water content QS Quality standard according to the proposed Groundwater Directive (QS) D2.4 – Guidance document on fate, transport and exposure for PMT’s 17 List of Abbreviations Qs Saturated soil water content R Retardation R2Coefficient of determination RBM River basin management RBSP River Basin Specific Pollutants (WFD) REACH Regulation concerning the Registration, Evaluation, Authorisation and Restriction of CHemicals SB-AP SimpleBox - Aquatic Persistence Dashboard SEAWAT Three-dimensional variable density groundwater flow and transport model based on MODFLOW and MT3DMS SFA Substance Flow Analysis SGW Soil-Groundwater SWI Soil-Water Interface SZ Saturated zone t Time UNZ Unsaturated zone V Velocity WFD Water Framework Directive, EU regulatory instrument for water management WHO World Health Organization of the United Nations WPC1 Water Protection Control Scenario 1 WPC2 Water Protection Control Scenario 2 WWTP Waste Water Treatment Plant x Distance D2.4 – Guidance document on fate, transport and exposure for PMT’s 18 1. Introduction Keymessages of the chapter Quantitative models for exposure assessment are needed to provide information on locations and times where monitoring data are not available, monitoring is not even possible (e.g. scenarios), for improved understanding of reasons for exposure and as basis to evaluate effectiveness of measures. Exposure of surface and groundwater to Persistent, Mobile and potentially Toxic substances (PM(T)) depends on use patterns and environmental behaviour of the chemicals. Emission, fate and transport models put driving factors in documented algorithms. This document explains the basic concepts of selected models designed to answer different specific questions and how they can best be used in combination in so-called “model trains”. The objective is to inform users and interested stakeholders on what needs to be considered for their application, what output and support they can provide and what not. Theoretical background is supported with good practice examples from the H2020 PROMISCES project. 1.1. Emission, fate and transport modelling for exposure assessment What is exposure assessment? Exposure assessment is the process of quantifying the dose of a contaminant (or a mixture of contaminants) that reaches a target organism. This is commonly followed by a second step, which quantifies the subsequent effect on the target organism. The total process is then referred to as “risk assessment”. In the context of human health, humans are the target organisms, with specific groups, such as infants or workers in factories, being the subject of study in certain cases. In accordance with the definitions set forth by FAO and WHO, exposure assessment is defined as “the evaluation of the probable intake of biological, chemical or physical agents via food, as well as exposure from other sources, if relevant”. In the context of assessments of aquatic ecological status, conducted in accordance with the Water Framework Directive, the specific target organisms are aquatic species (e.g. algae, macro-invertebrates, fish). The precise definition of “dose” depends on the subsequent effect assessment. It may, for instance, be the average intake of a contaminant via food, the “internal” concentration of the contaminant in vital organs, or the concentration in the surrounding environment (“external concentration” or “environmental exposure” for aquatic organisms in the water body). D2.4 – Guidance document on fate, transport and exposure for PMT’s 19 1. Introduction An exposure assessment requires a conceptual framework. It is necessary to carefully delineate the sequence of steps, from the source of the contaminant to the adverse effect on the target organism. This involves identifying sources of exposure, the behaviour of the contaminant in the environment, and the way in which the contaminant reaches the organism. The terms “Exposure Pathway” (the way the contaminant travels from the source to the target organism) and “Exposure Route” (the way the contaminant enters the target organism) are frequently employed. A commonly addressed aspect is also the uncertainty inherent in the exposure assessment. Exposure assessments can have a diagnostic objective. If a target organism has been identified as being exposed (via biomonitoring), the implementation of effective mitigation strategies requires the unravelling of the exposure routes, pathways and sources that caused exposure. Exposure assessments may also be conducted with a prognostic objective. Let us consider a scenario in which an industry wishes to start producing and using a new chemical. Will the intended use volume and use types lead to unacceptable exposure and subsequent human health risks? It is incumbent upon the exposure assessment to answer this question. Aim of modelling in exposure assessments Models are used in exposure assessments for two main reasons. The first reason is to complement monitoring data: it is often impractical or prohibitively expensive to monitor the environmental exposure at all times and along all exposure pathways and routes. In this context, modelling can assist in mapping the temporal and spatial variability of exposure, exposure pathways and exposure routes. Further, modelling can be employed to conduct hypothesis testing: can the known sources along the known pathways indeed be responsible for the measured exposure along the known exposure routes? If the answer is no, the underlying conceptual framework requires reconsideration. A second rationale for utilising models is to evaluate potential future developments. Will the proposed mitigation measure prove effective in achieving the required reduction in exposure? Or, as in the example mentioned in the previous section, will the intended introduction of a new chemical result in adverse human health effects? Or will climate change interfere with exposure pathways, and how? Emission modelling versus fate and transport modelling Fate and transport models calculate contaminant concentrations in the environment at a given location and at a given point in time. They are commonly “process-based”, meaning that they rely on mathematical equations reflecting our process understanding. Their input data comprise three main categories: 1. The characterization of the environment in which the contaminant under study needs to be simulated. This includes both the geometry of the environment and the fluxes of transporting media (air, water, fine particles). 2. The quantification of the emissions of the contaminant under study into the environment. 3. The definition of substance properties that control the fate and transport of the contaminant in question in the environment. This typically includes properties that determine the distribution of the contaminant over different media (air, water, organic matter, soil), D2.4 – Guidance document on fate, transport and exposure for PMT’s 20 1. Introduction as well as the degradation or transformation that the contaminant undergoes in different media. A fate and transport model may need to be expanded to calculate the required “dose” or environmental concentration. An example are so-called bio-accumulation models, which translate external concentrations into internal concentrations inside certain compartments/organs of the target organisms. The above almost automatically defines an emission model: a model that quantifies the emissions of a certain contaminant into the environmental compartments included in the fate and transport model. Emissions are expressed as mass per time. Depending on the complexity of the model chain, estimations of emissions can vary in spatial and temporal resolution and in considered compartments. Emission models can follow a wide spectrum of approaches, varying from completely data-driven modelling to almost fully process-based modelling. 1.2. Scope of the guidance document The scope of this document is to provide guidance for applications of models with specific focus on model trains for exposure assessment for persistent mobile and toxic chemicals (PMTs) as part of environmental risk assessment (specifically related to surface and groundwaters). This document explains the basic concepts of specific models and how these models can best be used in model trains in a tiered way. The term model train is used for a combination of standalone models to cover complex situations including several environmental spheres such as atmosphere, lithosphere, hydrosphere, biosphere, and anthroposphere, as well as their interfaces. Potential users are industries, utilities, environmental authorities, consultancies, and researchers. Further on, decision makers and regulators are addressed to increase the understanding of the potential policy implications and limitations of emission, fate and transport modelling in the context of environmental exposure assessment. This guidance document is neither a handbook on a specific model nor a toolbox for explaining different models to users. A toolbox for fate and transport models of PMTs in the environment is provided by the PROMICES-project in Deliverable D2.3 (Groot et al. 2025). The guidance provided by this document consists in assisting users to select appropriate methodological approaches (models and model trains) depending on which specific problem related to environmental exposure and related questions are posed, in order to identify and understand driving forces, potential future developments, and measures that could be taken to improve the situation in the light of European Green Deal including the Circular Economy Action Plan and of the Zero Pollution ambition of EU (EC COM(2019)640 final; EC COM(2020) 98 final; EC COM(2021) 400 final). The intention is to inform users and interested stakeholders on what needs to be considered for the application of different methods and on what can be expected and also what cannot be expected. In the guidance document we think more in terms of concepts rather than software/actual models. We focus on basic ideas, information requirements (data on regional circumstance and human activities, chemical properties, concentrations and loads in different media etc.), concepts of implementation, basic physical background, and comparison of different approaches to quantify processes by single models and on the role of single models in the context of a model train. Theoretical background is supported with good practice examples of model train applications from PROMISCES case studies. D2.4 – Guidance document on fate, transport and exposure for PMT’s 21 1. Introduction The structure of the document follows specific exposure routes. We present (i) “screening level” models for the assessment of regional exposure of groundwater from soil pollution and for the assessment of general exposure of air, soil and water at local, regional or global scales, with less detail as more targeted approaches, (ii) spatial and temporal explicit approaches for the identification of pollution plumes from contaminated soil in groundwater and (iii) model train applications for the catchment– river – river bank filtration – drinking water interactions. For the latter we focus on the large scale in a river basin and on a medium scale in an urban context. Table 1 gives an overview on the considered exposure types, related emission routes and considered fate and transport processes. D2.4 – Guidance document on fate, transport and exposure for PMT’s 22 1. Introduction Table 1.1.: Exposure, pollution sources/emissions, fate and transport considered in different model trains and related case studies. Considered exposure Considered compartments Fate and Transport Related case study Graphic Screening level Models Local, regional and global exposure of water, soil and air Local, regional and global emissions into water, soil and air Interaction of environmental spheres - Generic risk levels for soil for protection of groundwater Emissions into soils at regional scale Soil to groundwater - Soil – Groundwater Ground water Contaminated soil sites Plume development (time and space) Large-scale experiment (CS#6) and Tordera aquifer (CS#7) Catchment – river – river bank filtration – drinking water Surface water Terrestrial and man-made compartments in a large catchment Transport, retention, degradation in surface waters Upper Danube catchment (CS#2) Ground water & Drinking water Surface water impacted by terrestrial compartments Transport, sorption and degradation during bank filtration Vienna and Budapest (CS#2) Surface water Terrestrial and man-made compartments in an urban setting Transport, retention, degradation in surface waters Berlin (CS#1) D2.4 – Guidance document on fate, transport and exposure for PMT’s 23 1. Introduction 1.3. Basics of exposure assessment and environmental fate Chemicals are part of our life. Several hundred thousand are used in multiple applications in the European Union (Z. Wang et al. 2020) and as consequence may reach environmental spheres and water systems. Chemicals are used as pharmaceuticals, personal care products, pesticides/biocides or so-called industrial chemicals, which find application in manifold industrial processes or consumer products. Losses to the environment may occur throughout all stages of the life-cycle of products (i) from industrial production of chemicals, (ii) during finishing of consumer products, (iii) associated to consumptive use, (iv) from wear or ageing of products and materials and (V) during disposal from waste management (Figure 1.1). In the environment, these losses may lead to an exposure of aquatic life or humans via different emission pathways and exposure routes. The extent of the exposure depends on several factors determining the extent of emissions into the environment and the substances behaviour in the environment after their release. Emissions/releases into the environment in specific regions are driven by the amount of a chemical used (production plus import minus export) and the mode of use: private/industrial, use in closed systems, use resulting in inclusion into or onto an environmental matrix, non-dispersive or wide dispersive use. Figure 1.1.: Losses of chemicals to the environment via stages of the products life-cycle: (i) industrial production of chemicals, (ii) finishing of consumer products, (iii) consumptive use, (iv) wearing or aging of products and materials and (v) disposal from waste management (modified from Li et al. 2021) Once released into the environment, distribution and hence the exposure of organisms at a certain place depends on the chemical-physical properties of a substance. These determine the processes relevant for distribution as advection or adsorption/desorption to soil or sediments, sedimentation/resuspension with sediment, degradation as well as diffusion and volatilisation (Figure 1.2). Already the characterisation of chemicals as persistent and mobile (PM) or very persistent and very mobile (vPvM) substances indicates most important factors of environmental behaviour: the persistence and the mobility. Persistence describes the stability against degradation of a chemical under environmental conditions. While the mode of degradation/decay might differ between e.g. hydrolysis, photolysis or biodegradation and highly depends on the surrounding medium (e.g. soil, water, waste water treatment plants or soil), the shape of the degradation curve is typically considered to be a first order (exponential) decrease and can therefore be quantified by a first order rate constant/decay constant ( λ [d -1 ]) or a half-life time ( T1 2 [d]). Degradation may thus be described by the following D2.4 – Guidance document on fate, transport and exposure for PMT’s 24 1. Introduction formula, where C 0 is the initial substance concentration in a medium and C t the concentration after time t. T1 2 and λ are quantitatively related to each other and are characteristic numbers for specific substances in specific media. Typically, they are determined under standardised test conditions. The higher T1 2and the lower λis, the higher is the persistence of a substance. Ct=C0e−tλ (T1 2=ln2 λ)(1.1) A chemical is considered mobile in the environment, and more specifically in the water system, if it has a low affinity to be adsorbed to particles such as soil, sediments or suspended matter and is therefore transported by advection with the water flow. The adsorption in laboratory tests is frequently characterized by the octanol-water partition coefficient K OW [-], the organic carbon-water partition coefficient K OC [m 3 /kg] or the linear adsorption isotherm for soil or sediment Kd[m3/kg], where: K OW = C O /C W ; (C O ...concentration of a substance in octanol, C W ...concentration in water) KOC = COC/CW(COC. ..concentration of a substance in organic carbon) Kd= KOC x fOC (fOC. . .fraction of organic carbon in soil) In practice, K OW and K OC are often used as logK OW and logK OC , where values below 3 indicate chemicals with a low tendency to adsorb onto organic carbon (contained in soils and sediments) and therefore with a high mobility in the environment, whereas values above 4 indicate chemicals with a high tendency to adsorb to organic carbon and therefore a low mobility in the environment. While T1 2 and λ as well as K OW , K OC and K d are the most important parameters to characterize environmental behavior of a substance and are essential in many applications of emission, fate and transport modelling of environmental exposure, there are many other physico-chemical properties that might be used for specific cases. Among them, molecular weight [g.mol -1 ], water solubility [mg.l -1 ], vapour pressure [Pa] and boiling point [°C] are worth being mentioned. A specific focus on methodologies from the field of chemo-informatics to assess substance properties relevant for environmental behaviour is given within the PROMISCES-project in deliverable D2.1 “Toolbox improved in silico models for PM(T) properties identification” (Sosnowska et al. 2024). D2.4 – Guidance document on fate, transport and exposure for PMT’s 25 2. Screening level models which are inserted as ranges (Table 2.1), so that the relatively large uncertainty in aquatic persistence for these substances can be explained with the sensitivity plots that can be displayed in the SB-AP Dashboard (Meesters, J.A.J. 2024) (Figure 2.2). NFBS 6:2 FTOH FTSAAm Figure 2.2.: Overview of sensitivity plots for substances with NFBS, 6:2 FTOH and 6:2 FTAB for aquatic persistence (s) vs water degradation rate constant (log s -1 ) and log K OW For substance NFBS the uncertainty in the calculated aquatic persistence of dissolved species as well as the sum of dissolved and sorbed species is due to the rate constant for degradation in water. The respective plots show a clear linear decrease of aquatic persistence of NFBS with increasing water degradation rate constant, whereas the log K OW plots only displays noise. For D2.4 – Guidance document on fate, transport and exposure for PMT’s 32 2. Screening level models 6:2 FTOH and FTSAAm, there is a difference between aquatic persistence of the dissolved and sorbed species. The plots display clear decreases in aquatic persistence of dissolved species with increasing K OW , whereas for the sum of dissolved and sorbed species the aquatic persistence increases with increasing K OW . Within the SimpleBox model, where the SB-AP Dashboard is plugged into, the K OW determines how strong substances sorb to the suspended matter, so that substance in dissolved species are removed from the water column. However, the SB model also calculates that sorbed substance mass is not prone for degradation in water, so that increasing K OW leads to more sorption and hence less degradation. Such a simulation of chemical fate by SB can be seen in the sensitivity plots for 6:2 FTOH and FTSAAm, as (i) stronger K OW decreases their aquatic persistence of dissolved species only, (ii) the sum of dissolved and sorbed species increases with increasing KOW and (iii) an increasing water degradation rate constant affects the aquatic persistence of the sum of dissolved and sorbed species to a minor extent only. The SB model also simulates removal of substance out of the water column via sorption to suspended matter that settles to sediment layer at the bottom. Nonetheless, the aquatic persistence increases with increasing K OW for 6:2 FTOH and 6:2 FTAB. It thus appears that the impact of blocking their degradation due to sorption is simulated to be stronger than the removal via sorption with settling suspended matter. The substances cC604 and 6:2 FTAB are calculated to be the most persistent in aquatic environments. Their aquatic persistence exceeds the vPvM index on a regional scale with a factor 4.6 (851 – 867 days vs 187 days), whereas on a continental scale this is a factor 6.5 (1270 – 1320 days vs 203 days) and even a factor of 270 on a global scale (43,686 - 58,000 days vs 215 days). For these two substances it can thus be concluded that their persistence in aquatic environments lasts for decades on a global level. Key messages for the application of the SimpleBox - Aquatic Persistence Dashboard The combined impact of (P)ersistent and (M)obile properties of a substance can be quantified with aquatic persistence as calculated with the SB-AP Dashboard. This is useful in screening level substance evaluation by comparing the aquatic persistence 1. between different substances, 2. against PM or vPvM indexes or 3. between a substance’ dissolved and/or sorbed form. The impact of uncertain substance properties can be quantified and explained with sensitivity plots. D2.4 – Guidance document on fate, transport and exposure for PMT’s 33 2. Screening level models 2.3. Deriving generic risk limits for soils 2.3.1. Model explanation Soil contamination can lead to the degradation of groundwater quality in the long term, especially if the contaminants are persistent. Establishing risk-based soil quality criteria is a common approach for policy makers to deal with the effects of soil contamination on health and the ecosystem (Swartjes 2011). Most of these criteria are based on direct exposure however, whereas indirect effects, for instance through drinking water quality deterioration, stemming from soil pollution, are not part of the derivation of generic criteria. The following approach proposes a method to derive generic risk limits for soil (in situ or soil that is to be used, or ‘applied’ elsewhere) that can be applied on a national or a regional scale to underpin soil quality criteria. These risk limits based on leaching can be used alongside more traditional risk limits based on direct exposure. Contaminants in soils are partitioned into an adsorbed phase and an aqueous phase. Under net infiltration, the contaminant will be transported towards the groundwater. In the simplest form, the partitioning is expressed as a partitioning constant Kp. Kp=Csoil Caqueous (2.2) With C soil being the total concentration in soil and C aqueous being the dissolved concentration in the aqueous phase. The leaching of a contaminant is dependent on aspects that influence the partitioning of the (v)P(v)Ms and the transport of soil moisture. Partitioning is amongst others affected by the chemical properties of the (v)P(v)Ms as well as the chemical properties of the soil. The transport of soil moisture is a function of the hydrological balance and soil hydraulic properties. Generic leaching risk limits (GLRL) for soil or sediment (that is to be re-used on land) can be derived using reactive transport modelling. These risk limits can aid in making decisions on applying soil or sediment in certain locations. This is especially important in the case of the re-use of material of a different origin than the location of application and as well in the case of vulnerable hydrogeological circumstances (e.g. shallow groundwater table, drinking water wells nearby). Central to the derivation of generic leaching risk limits are standard scenarios. A standard scenario refers to the collection of (v)P(v)M properties, soil properties, both chemical and hydrological, climate data, starting conditions with regards to soil contamination and endpoints which are subsequently used to derive GLRL. (v)P(v)M properties The properties of the (v)P(v)Ms are of importance as these determine potential physicochemical transformations and interactions with soil components. Upon entering soils, compounds can undergo adsorption-desorption interactions with surfaces, degrade, volatilize or form precipitates. D2.4 – Guidance document on fate, transport and exposure for PMT’s 34 2. Screening level models Such reactions influence solute transport. Whilst (v)P(v)Ms are per definition persistent and mobile, which implies that they do not degrade readily and that adsorption interactions are limited, such processes need to be accounted as persistence and mobility are relative, and substantial differences may be present between substances. To derive a GLRL, it is necessary to consider which physiochemical properties and interactions are relevant to describe solute transport accurately. Adsorption interactions should always be considered as part of the derivation as these can vary depending on the type of substance and soil, leading to differences in the derived GLRL’s. The manner in which adsorption can be accounted depends on the modelling software used. It is unlikely that the requisite thermodynamic data to implement a mechanistic model is available for many (v)P(v)Ms and therefore only empirical models are suggested. Furthermore, for the derivation of GLRL, it is assumed that (v)P(v)M concentrations are well below the adsorption capacity of soils and that empirical models are therefore applicable. This holds almost without exception for soil contamination stemming from diffuse sources. When deriving and applying GLRL to be used in situations with higher contamination levels, e.g. near a point source, this condition should be explicitly verified. The model used per (v)P(v)M will vary depending on the information available. Ideally, a Freundlich isotherm that accounts for soil properties based on a multiple regression analysis is implemented. If this information is not available, a linear adsorption isotherm (K d ) that considers soil properties, or a K OC in the case of hydrophobic compounds, can be used. When using empirical models that account for soil properties, it is important that properties of the soil being modelled fall within the range of properties used to derive the relationship between the isotherm and properties. Furthermore, it is important that the concentration range of the (v)P(v)Ms being modelled, falls within the concentration range for which the empirical relationship was derived. If only a linear adsorption model is available for a (v)P(v)M, then it may be necessary to identify a K d from literature and assign a generic decrease in the K d with soil depth, to reflect the decrease in reactive surfaces (e.g. organic matter and clay content). In this case, a conservative approach is suggested, for instance by selecting a K d from the lower end (e.g. a 5 th or 10 th percentile) of K d values reported in literature for simulation purposes. It should be noted that selecting a specific K d value from a distribution of curated values is in the end a matter of policy. If K d values are not present in literature, then a value derived from a QSAR may be applicable. Soil properties Reactive surfaces in soils and soil solution chemistry influence (v)P(v)M transport. Variations in these factors therefore lead to differences in solute transport between soils. The hydraulic properties of soils influence waterflow and therefore well-draining, coarse textured soils, or soils characterized by macropores may facilitate contaminant transport to groundwater. Due to the diversity in soil properties it is appropriate to derive GLRL using soil profiles representative of major soil types present in a state or country. If these profiles are not available, local soil experts should be consulted to identify representative soil profiles. For each soil profile, the average groundwater depth and relevant soil properties should be determined. The soil properties should be determined at intervals that capture the vertical heterogeneity of the D2.4 – Guidance document on fate, transport and exposure for PMT’s 35 2. Screening level models Table 2.2.: Potentially relevant soil chemical and physical parameters to be collected for the purpose of deriving generic risk limits for leaching Soil property type Parameters Chemical pH, organic carbon, soil texture, Fe-oxide, Al-oxide, CEC, AEC Physical Ks, θs,θr, n, ρ soil profile. A list of potentially relevant soil chemical and physical properties is presented in Table 4. Note that because the depth of the groundwater table varies the depth to which soil property data needs to be collected may vary per soil profile. At the very least, the data needed to parameterize a single-porosity waterflow model is required with regards to soil hydraulic properties. If these parameters are not determined then the European pedotransfer functions (Szabó et al. 2021) can be used as an alternative until measurements have been conducted. Climate (v)P(v)M transport from the unsaturated to the saturated zone of soils is assumed to occur primarily by infiltrated precipitation. Downwards transport of water through a soil profile generally (i.e. in the absence of preferential flow) occurs when the field capacity is exceeded. Increased water transport implies an increased capacity for solute transport and therefore the balance between precipitation and evapotranspiration is an important factor to consider when deriving GLRL. The balance between precipitation and evapotranspiration varies within and between countries. It is therefore necessary to use average daily precipitation datasets that are representative for the regions which the soil profiles represent. Ideally, data on the last 30 years of net precipitation input is used for the derivation of GLRL. Initial conditions, boundary conditions and endpoints The initial conditions refer to the depth of soil contaminated, the characteristics of contamination, the concentration of the contaminant and the groundwater depth. Together with policy makers, it is necessary to specify what portion of a soil profile is contaminated (e.g. top meter) and to characterize the distribution of contamination (e.g. homogenous or heterogenous) in the standard scenario. Furthermore, for simplicity, using a fixed groundwater depth as the lower hydrological boundary condition is suggested. Identical initial conditions should be used for each of the soil profiles for which GLRL are to be derived. Endpoints are the points of compliance and associated environmental criteria. The point of compliance for GLRL is the upper meter of groundwater. Furthermore, the GLRL should be determined on the basis of peak concentrations reached with simulations. The environmental criteria the GLRL are based on human and/or ecological risk limits for the (v)P(v)Ms. The selection of appropriate risk limits in groundwater is a matter for policy. D2.4 – Guidance document on fate, transport and exposure for PMT’s 36 2. Screening level models 2.3.2. Application cases of generic risk limits for soils and dredged material (sediment) In this study, environmental quality criteria for PFAS in ground and surface water, e.g. based on WFD quality targets or drinking water protection, are the starting point for the calculation of corresponding concentrations in the soil or dredged material (sediment) to be used for the derivation of generic risk limits. Conceptual models have been drawn up for two main variants: 1. Groundwater. This main variant assumes that PFAS that are carried along with infiltrating rainwater end up completely in the groundwater. The standard used for groundwater in the baseline scenario is the health-based guideline value for drinking water. In this example a guidance value for PFAS of 4.4 ng/l PFOA equivalents was used (Schepens et al. 2023), a value that is based on the EFSA et al. (2020) health-based guidance. Three points or planes of compliance (POC) are defined (see Figure 2.3): POC 0, which is the point of infiltration in the unsaturated zone. POC1, an intermediate point at the phreatic plane. This point is located directly below the application and marks the transition from (the modelling of) vertical transport to horizontal transport). Finally, POC2 is located at a distance of 25-year transit time (based on groundwater flow) in the aquifer starting from POC0. Note that the selection of a certain transit time is a policy decision. 2. Surface water. This main variant assumes a relatively shallow aquifer in which complete mixing takes place of PFAS originating from the top soil layer (Figure 2.4). Note, that this is typical for many areas in the Western part of the Netherlands (‘polders’), but in other areas combined systems of aquifers and surface water may run much deeper. An important outcome of this more or less typical ‘Dutch variant’ is that the dilution is relatively limited in comparison the other variant. The standard for surface water in the initial scenario is the current surface water standard in the Water Framework directive for PFOS (0.65 ng/l) and a similarly derived value for PFOA (48 ng/l, Verbruggen et al. 2017. The groundwater variant is modelled using a Hydrus and Modflow modelling train (see chapter 3) whereas the surface water scenario is calculated using a simple dilution calculation based on the assumption of instantaneous mixing in the aquifer and a flux through POC2 that is equal to the net infiltration. The risk limits are intended to substantiate generic standards that must apply to many different situations. It is therefore not operationally practical to establish criteria for a large number of different variants and scenarios. Depending on the spatial scale and variability of environmental conditions an appropriate number of generic risk limits may be derived. The variance of risk limits calculated for different cases can be used to aid in deciding the required amount of different values in relation to the scale and variability of the use case. As an example, a single baseline scenario has been chosen on the scale of the Netherlands for both of the above-mentioned main variants with the following characteristics: • Based on an average concentration in the plane compliance POC 2, as opposed to setting a maximum concentration (only applicable to groundwater variant); •Thickness of applied soil (soil that is to be re-used) layer of 1 meter; D2.4 – Guidance document on fate, transport and exposure for PMT’s 37 2. Screening level models Figure 2.3.: Conceptual model for the groundwater variant Figure 2.4.: Conceptual model for the surface water variant D2.4 – Guidance document on fate, transport and exposure for PMT’s 38 2. Screening level models •Surface area of soil application of 25 m ×25 m; •Application of soil above the groundwater table (as opposed to below the surface); • Groundwater level 1 meter below ground level (representative for large parts of The Netherlands); •Not taking background concentrations in soil and groundwater into account; • 10th percentile of the sorption coefficients from previous leaching research (Wintersen et al. 2020); The criteria at POC2 in the groundwater variant are the drinking water guidance value of 4.4 ng/l for PFOA and 2.2 ng/l for PFOS and for surface water biota-based (human fish consumption) WFD(-type) criteria of 0.65 ng/l and 48 ng/l (Verbruggen et al. 2017) for PFOS and PFOA respectively. Table 2.3.: Derived GLRLs for PFOS and PFOA for soil/sediment from the Dutch case study for different points or planes of compliance (POC) in groundwater and surface water. GLRLs for soil/sediment [µg/kg ds] Substance Groundwater Surface water PFOS 48 0.14 PFOA 37 3.4 The calculated risk limits for leaching associated with the baseline scenarios for the two main variants are shown in Table 2.3. These risk limits can be used to assess both in situ soil and soil that is to be applied (re-used) elsewhere. These values should be considered as the result of a first proof of concept for the Dutch situation. Depending on further (policy) choices and local variations, different values can apply. The calculated risk limits expressed as concentrations in soil/dredged sediment based on the surface water variant are lower that the values calculated for the groundwater variant. This is caused by the combination of stricter risk limits in surface water and the smaller amount of dilution in this scenario. Key messages for deriving generic risk levels for soils The relatively high mobility of some PFAS combined with low environmental quality criteria for ground and surface water can necessitate the derivation and enforcement of criteria that are to be applied on soil and sediment that is to be applied on land; This case study provides generic guidelines for deriving such criteria. Regional variations in geo(hydro)logical circumstances, leading to variations in sorption, should always be considered when deriving such criteria; As an example, generic criteria for Dutch soils and sediments are presented, based on a parameter set that is typical for Dutch situations (shallow groundwater table and close proximity to surface water). D2.4 – Guidance document on fate, transport and exposure for PMT’s 39 3. Soil - Groundwater interaction Key messages of exposure assessment in the soil groundwater continuum The model train developed is efficient to simulate PFAS fate and transport in the soil-groundwater continuum from semi-real to aqueous film forming foams (AFFF) contaminated site scales The simulation results highlight the key role of the unsaturated zone in the PFAS transport in the soil-groundwater continuum to correctly predict the extension of PFAS plume. Capillary fringe displacement plays a key role in PFAS remobilisation by controlling water flow rate in the unsaturated zone and modifying PFAS sorption onto air-water interface. Numerous processes are involved in PFAS transport in the soil-groundwater continuum. Most of them are interrelated. 3.1. General aspects In many soils, measured PFAS are orders of magnitude greater than the health advisory level. A quantitative understanding of the fate of PFAS in soils and to other environmental compartments such as groundwater is therefore mandatory to develop effective remediation or mitigation strategies. PFAS migration in porous media is one of the most challenging as it needs to account for variably saturated media (soil and unsaturated zone) as well as fully saturated media (groundwater). Modelling the fate and transport of PFAS in this continuum are challenging due to the following several issues: • Lack of knowledge on the physical, chemical and biological processes controlling fate and transport of PFAS in the soil-groundwater (SGW) continuum, • Difficulty to assess spatio-temporal patterns of heterogeneities in SGW continuum and their consequences on PFAS reactive transport, • Diversity of the numerical formalisms, from empirical ones to physically based, preventing to identify the most appropriate to simulate the processes controlling PFAS reactive transport, • Uncertainties concerning the validity of upscaling simulation results and modelling tools from laboratory scale to contaminated site scale, D2.4 – Guidance document on fate, transport and exposure for PMT’s 40 3. Soil - Groundwater interaction • Lack of modelling tools to build 3 dimensions (3D) PFAS reactive transport model in transient time conditions, • Existence of a broad range of PFAS with contrasted molecular properties, implying to develop modelling tools that can be applied to all, or at least, most of these compounds, • Spatial/temporal variability and uncertainty of PFAS monitoring at field scale, often preventing to conduct robust model calibration, • Scarcity of the data used to estimate PFAS emissions at the top of the SGW continuum, questioning the boundary conditions selected as input data in the modelling tools, • The broad range of PFAS concentration encountered from the hot spot of contamination (mg/L) to the fringe of the plume where the concentration is lower (µg/L), Up to now, PFOA and PFOS are the most studied PFAS in the literature as they are the most used and older PFAS used. In addition, thresholds values have been defined in drinking water directives (0.1 µg/L for each) or EPA is setting enforceable Maximum Contaminant Levels at 4.0 ng/l for PFOA and PFOS, individually. Although a growing number of research works have been conducted to simulate reactive transport of numerous PFAS in SGW continuum, developing a robust and efficient modelling tools whom results can be used by various stakeholders, policy makers or operating agents dealing with problems of PFAS contamination in groundwater is still a complex task not clearly defined up to now. To efficiently develop modelling tools to simulate fate and transport of PFAS in SGW continuum several steps must be conducted. Note that several of these steps are also found in the modelling framework developed for “screening level” models (see section 2) and model train applications for the catchment – river – riverbank filtration – drinking water interactions (see section 4). These steps are the following: • Conceptualize a robust and comprehensive conceptual model describing the key processes controlling fate and transport of the studied PFAS in the studied SGW continuum, • Build a model train combining several numerical tools/software that are efficient to simulate the processes identified during the first step, • Selecting the most appropriate mathematical formalisms to simulate the processes controlling fate and transport of the studied PFAS in the studied SGW continuum •Perform an efficient model calibration using a dedicated dataset, • Analyse modelling results for each simulation to characterize and rank the main processes involved in reactive transport of the studied PFAS in the studied SGW continuum, • Estimate origins (conceptual, numerical, calibration, etc.) of discrepancy between simulated and measured results, • Identify the best improvements steps to update the model for improving its efficiency to simulate and forecast the reactive transport of the studied PFAS in the studied SGW system. Considering the two first steps aforementioned, a solid knowledge on the processes controlling PFAS migration in SGW continuum is required. A brief review of the key concepts to simulate PFAS fate and transport in SGW continuum is provided in Annex B.1 D2.4 – Guidance document on fate, transport and exposure for PMT’s 41 3. Soil - Groundwater interaction interactions between groundwater and surface water. PFAS contamination has been induced by repeated used of AFFF for fire-fighting training activities. Field campaign measurements have been done to characterize groundwater flow, variably saturated flow in the unsaturated zone and PFAS contamination. For each case study, an example on the use of the model train is provided. The main steps aforementioned for building a model to simulate PFAS fate and transport in the SGW continuum are explained below. More details on the PMP experiments and CS#7 are given in the Annexes (B.2 and C, respectively). 3.4. Application for the PluriMetric Pilot experiment This section describes the model of the PluriMetric Pilot experiment that is described in detail in Annex B.2. 3.4.1. Conceptual model Based on the interpretation of the results issued from the PMP experiment, a conceptual model explaining PFAS fate and transport during the PMP experiment has been built. The conceptual model considers four processes: i) water flow in both unsaturated and saturated zone controlled by different type of boundary conditions, ii) advection-dispersion, iii) sorption reactions onto soil-water interface (SWI) in both unsaturated and iv) saturated zone and sorption reactions onto air-water interface (AWI) in unsaturated zone. The magnitude of these processes, hence their role in controlling PFAS fate and transport, changes with distance to the AFFF infiltration area, named infiltration square, along the water flow path and time. In consequence, several hypotheses have been assumed to explain the spatio-temporal pattern in PFAS migration. These hypotheses are the following: • At the beginning of the AFFF infiltration stage, pore volume in both unsaturated and saturated zone below the infiltration square has decreased compared to the values defined for the uncontaminated geomedia. This change in pore volume is assumed to be induced by foam formation. • By sorbing onto soil-water and air-water interfaces, PFAS have modified the hydraulic properties of the geomedia, increasing water flow rate in both unsaturated and saturated zone below the infiltration square. • Changes in pore volume and hydraulic properties have been stable during all the PMP experiment since starting AFFF infiltration • The first centimetres of the unsaturated zone seems less impacted by foam formation, hence hydraulic properties and pore volume remained like the ones measured in the uncontaminated geomedia. • At the top of the PMP, evaporation has impacted both water flow rate and has induced an increase of PFAS concentration in pore water. D2.4 – Guidance document on fate, transport and exposure for PMT’s 48 3. Soil - Groundwater interaction • Water table depth as well as boundary conditions controlling water flow in the saturated zone remained constant during all the experiment • Sorption reactions on soil-water interface have occurred in both unsaturated and saturated zones for the three main PFAS found in the AFFF solution (6:2 FTSA, 6:2 FTAB, 6:2 FTSAam) • Sorption reactions on air-water interface have occurred for the three main PFAS in the unsaturated zone. The magnitude of this reaction seems to decrease along the depth of the unsaturated zone. More the geomedia was unsaturated, greater the sorption reactions onto AWI has been. • The three main PFAS in AFFF can be classified according to their concentration in pore solution as following: 6:2 FTSA » 6:2 FTAB > 6:2 FTSAam. • The three main PFAS were not affected by degradation reactions during the PMP experiment • No competition between the three main PFAS was expected for the sorption sites either onto SWI or AWI. 3.4.2. Numerical model set up Based on the conceptual model, a model train combining: i) a groundwater flow model developed using MODFLOW software, ii) 1D-vertical model for simulating water flow in the unsaturated zone developed using 1D-Hydrus variably water flow module, iii) a groundwater solute transport model developed using MT3DMS and iv) a 1D-vertical model for simulating solute transport in the unsaturated zone developed using 1D-Hydrus solute transport module. The numerical scheme used to couple the software is the option 2 (see the paragraph ’Coupling scheme’ in section 3.2.2). For each model of the model train, 5 main steps have been performed. They are described below. Spatial and temporal discretization The first step has been to define geometry and meshing of the groundwater flow model. The PMP has been described as a full 3D space of 5.2 m long, 3.6 m width and 3.0 m depth. This domain is discretized in 7020 square cells, each square cell has a side of 0.2 m (0.2 long, 0.2 width and 0.2 depth), except for the first layer of cell whom thickness is greater (1.2 m). The number of layers in the groundwater flow model is equal to 10. The first layer corresponds to the unsaturated zone in the PMP experiment. In this first layer, water flow is simulated by a 1D-vertical model built using 1D-Hydrus instead of the recharge modules available in MODFLOW. The thickness of the first layer has been defined to be sure that the bottom of the 1D-vertical model is below the deepest possible water table level that can occur during the simulations. To keep the model as simple as needed, a 1D-vertical model was not built for each cell of the first layer of the groundwater flow model. Two groups of cells have been defined, the cells located in the square where AFFF has been infiltrated and the cells outside of the AFFF infiltration square. In each group, hydrodynamic, solute transport and reactive properties as well as the associated boundary conditions are assumed identical for all the cells. Therefore, D2.4 – Guidance document on fate, transport and exposure for PMT’s 49 3. Soil - Groundwater interaction only two 1D-vertical models were built. For these two 1D-vertical models, the simulated domain is divided in 121 nodes, hence cell thickness is equal to 0.01 m. For the 1D-verical model built to simulate water flow in the AFFF-infiltration square, the cells have been distributed between 6 layers to simulate heterogeneities of the hydraulic properties as well as the decrease of PFAS sorption onto AWI along depth. As the thickness of each layer cannot been characterized, these values have been determined by calibration. The same geometry and meshing were used for the groundwater solute transport model and the 1D-vertical models developed for simulating solute transport. The time length of the PMP experiment (205 days) has been divided into 4920 stress periods. Each stress period has therefore a length of 1 h. For each stress period, boundary conditions are defined (see below). Boundary and initial conditions In the PMP, water in the saturated zone flows from left side to right side. On left side, constant head boundary is prescribed. On right side, a pumping well is located. The pumping rate is kept constant as it has been done during all the PMP experiment. For the unsaturated zone, transient boundary conditions are considered. More precisely, recharge, potential evaporation and solute fluxes are simulated at the top of the 1D-vertical models simulating water flow and solute transport in the cells of the AFFF infiltration square. The amount of infiltrated water (with and without AFFF) during the PMP experiment is simulated. For potential evaporation, mean values have been calculated for each hour based on the data collect periodically on air temperature and air humidity at the top of the PMP during the experiment. The same pattern of evaporation has been reproduced for each simulated day. To simulate solute fluxes, transient conditions are also considered. Prior and after AFFF infiltration, the water infiltrated is PFAS-free. During the infiltration stage, the infiltrated water contains non-reactive tracer compounds (bromide) and the three main PFAS found in the AFFF (6:2 FTSA, 6:2 FTAB and 6:2 FTSAam). The concentrations of these compounds are assumed equal to the mean values calculated based on the time series measurements performed on the input solution. For the 1D-vertical model used for the cells outside the infiltration square, only potential evaporation has been fixed as boundary conditions. Concerning initial conditions, the water content profile at the starting of the PMP experiment cannot be measured. In order to obtain steady-state conditions prior to standard the simulations of the PMP experiment, a prior modelling stage was simulated. During this stage with a time length of 70 days, a weekly infiltration of 120 L of PFAS free water is simulated in the cells of the infiltration square. Evaporation is also simulated considering the same values that during the simulations of the PMP experiment. Hydraulic properties In the two 1D-vertical models built, the Van Genuchten-Mualem equations are used to predict changes in hydraulic properties according to changes in pressure head. The values for these hydraulic functions are defined for each layer of the model. For the 1D-model developed for the cells located in the infiltration square, these values have been calibrated as they cannot be measured directly. For the 1D-model developed for the remaining surface of the PMP, the values were calculated using the distribution of the granulometric fractions in the geomedia. To avoid any discrepancy between the 1D-vertical models and the groundwater flow model, the D2.4 – Guidance document on fate, transport and exposure for PMT’s 50 3. Soil - Groundwater interaction same hydraulic properties in saturated conditions are considered in the cells at the bottom of the former model and in the cells of the latter. No longitudinal neither transversal anisotropy is assumed. Hence, values of longitudinal and transversal saturated permeability are equal. Specific yield is kept similar in all the layer and set equal to value determined for sandy materials (0.353). Solute transport properties The apparent density is fixed based on measurement and set equal to 1900 kg/m 3 . The longitudinal dispersion assumed in each layer of the 1D-vertical model developed for the infiltration square is calibrated (see below). For the 1D-vertical domain for the remaining part of the unsaturated zone, the longitudinal dispersion is assumed equal in the 6 layers and is set equal to 10. This value is similar to the ones used for sandy materials. The same values of longitudinal dispersion are set identical in the deeper layer of the 1D-vertical models (Layer 6) and the saturated layers in the groundwater solute transport model. In the groundwater solute transport model, longitudinal and transversal dispersion are set identical. Diffusion does not affect PFAS fate and migration during the PMP experiment, hence diffusion and tortuosity coefficients are fixed to 0. Reactive properties Prior experiments at laboratory scale have demonstrated that the sorption reactions of 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam in the geomedia onto SWI and/or AWI are not linear and not ideal (data do not show). Furthermore, these experiments illustrated that sorption reactions on SWI and AWI occur for the three PFAS. A non-equilibrium chemical approach accounting for two-site adsorption model is used to simulate PFAS sorption reaction in the unsaturated zone below the infiltration square. This approach approximates sorption reactions onto SWI and AWI in the unsaturated zone without distinguished them. The Langmuir isotherm is used to simulate instantaneous sorption reaction while a first order rate sorption equation is used for the kinetically controlled sorption reaction. The proportion of the two groups as well as the values for the Langmuir isotherm and the rate of the kinetically controlled sorption equation have been calibrated for each layer of the 1D-vertical model devoted to the unsaturated zone below the infiltration square. For the groundwater solute transport model located below and downstream of the infiltration square, the Langmuir isotherm is used to simulate sorption of PFAS onto SWI. The parameters values for cells below the infiltration square are calibrated. 3.4.3. Calibration As mentioned above, values of several model parameters have been calibrated. A calibration approach relying on fitting iteratively several objective functions (see below), minimizing the difference between simulated and measured values, was used to define the values of the model parameters. The metrics used to characterize the discrepancy between measurements and simulations are the normalized root mean square error (nRMSE) and the correlation coefficient D2.4 – Guidance document on fate, transport and exposure for PMT’s 51 3. Soil - Groundwater interaction (R 2 ). Furthermore, calibration was conducted by assuming constraints. One of them is that the a priori distribution of values for model parameter is narrowed based on literature knowledge. Spatial changes in model parameter values are also constrained based on the interpretations of the results of the PMP experiment. Last, interrelation rules between model parameters were also set. The modelling framework mentioned in section 3.2.3 was used, hence four steps was considered during the calibration. The first calibration step was dedicated to estimate values of pore volume, the hydraulic properties and the longitudinal dispersivity of the 1D-vertical model (both variably water flow and solute transport built using 1D-Hydrus) for the cells located in the unsaturated zone below the infiltration square were fitted first. The objective function relies on the Br concentration measured during the experiment for the three sampling points located to the unsaturated zone. Based on the interpretations of the PMP experiment (see Annex B.2), the pore volume decreases along depth in the unsaturated zone. Hence, the values of pore volume attributed to an upper layer cannot be lower than the one used for a deeper layer in the 1D-vertical model. For the permeability in saturated conditions as well as the parameters of the Van Genuchten and Mualem functions, the values are assumed similar all along the unsaturated zone, except in the first layer close to the surface where the pore network is assumed to be disturbed during the filling of the PMP. The second calibration step had focused on estimating the hydraulic properties and the longitudinal dispersivity for the cells located below and downstream the infiltration square in the saturated zone. Similarly, to the step 1, the objective function is to fit the Br concentrations measured during the experiment in the sampling point located below and downstream the infiltration square. As previously mentioned (see above), the model parameters in the cells of the model at the top of the saturated zone below the infiltration square was assumed to be identical to the model parameters at the bottom of the 1D-vertical model to avoid inconsistency in the capillary fringe. Furthermore, the impacts of AFFF infiltration on hydraulic and solute transport properties are assumed to be fading drastically according to the distance from the infiltration square. Therefore, for each model parameter, the range of values used for cells not below the infiltration square were assumed to be narrow and centered on the mean values estimated for the uncontaminated geomedia. As the geomedia is assumed isotropic, the same values were used for longitudinal and transversal dispersivities in the saturated zone. The third calibration step was devoted to estimate the values for the thickness of the 6 layers assumed in the 1D-vertical models, the distribution of the two sorption site groups, the partition coefficient of the Langmuir equation, the maximum amount of sorbing site available in the Langmuir equation and the rate of first order kinetically-controlled sorption reactions in the unsaturated zone. As sorption reactions onto the SWI and the AWI were not differentiated during the simulations, the model parameters for the Langmuir equation and the first order rate equations are combinations of the independent values for each one of the sorption reactions either on the SWI or the AWI. This calibration step was conducted independently for the three main PFAS present in the diluted AFFF solution (6:2 FTSA, 6:2 FTAB and 6:2 FTSAam) as no interactions between the three compounds were assumed. For each of the three compounds, the data used were the measured concentrations measured during the experiment in the three sampling points located below the unsaturated zone. As observed by interpreting the measured results, the intensity of the sorption reactions onto the AWI decreased with depth, i.e. more PFAS are sorbed onto the AWI in the upper part of the unsaturated zone below the injection D2.4 – Guidance document on fate, transport and exposure for PMT’s 52 3. Soil - Groundwater interaction square due to the higher surface of the AWI in the zone of the PMP, this latter resulting from the lower water content. The model parameters describing the sorption reactions onto SWI in the saturated zone were fitted during the last calibration step. As in the third calibration step, this step was performed independently for each one of the three main PFAS present in the diluted-AFFF solution. For each compound, data used are the concentrations measured during the experiment in the sampling points located below and downstream the infiltration square in the unsaturated zone. To avoid discrepancy between the 1D-vertical model and the groundwater solute transport model dedicated below the infiltration square, the same values for the parameters of the Langmuir isotherm were used for cells at the bottom of the 1D-vertical model and the cells at the top of the groundwater solute transport model. The geomedia is assumed to exhibit low heterogeneity. In consequence, sorption reactions onto the SWI are expected to be identical below and downstream from the infiltration square. Therefore, the same values for the model parameters mentioned above are used for all the cells in the saturated zone below and downstream the infiltration square, except for the cells at top of the groundwater solute transport model. The fitted values for hydraulic properties, solute transport properties and sorption reaction can be found in the Annex B.3. 3.4.4. Results The main demonstration issued from the modelling activities on CS#6 is that the model train was able to reproduce accurately the measured spatial and temporal changes in concentration of 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam during the PMP experiment, even if measured PFAS concentrations were overestimated or underestimated by the model at different sampling points (see Figure 3.2 and Figure 3.3). The changes in Br concentrations were also correctly reproduced (Annex B.4). As an illustrative purpose, the modelling results obtained along the main water flow path, below and downstream the infiltration square, are discussed. The model goodness-of-fit illustrates that the model developed using the model train for SGW continuum developed in the PROMISCES project is able to simulate the main physical and chemical processes controlling the fate of 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam during the PMP experiment. The comparison between modelled and measured concentrations for both non-reactive tracer and PFAS allows to validate or not the hypotheses of the conceptual model, improving our understanding of the main processes controlling PFAS fate and transport in the simulated domain. By demonstrating the high goodness-of-fit of the model to simulate spatial and temporal changes in non-reactive tracer concentrations in pore solution, three main hypotheses of the conceptual model can be validated: • Evaporation at the top of the PMP has impacted water flow and solute transport during the experiment. This process explains that the non-reactive tracer concentrations in the pore solution rose above the concentration in the infiltrated AFFF solution. These results highlight that an accurate description of the boundary conditions at the top of the simulated domain is a key step to correctly simulate the fate and transport of the PFAS. Considering a very long residence time for some reactive PFAS such as PFOS and D2.4 – Guidance document on fate, transport and exposure for PMT’s 53 3. Soil - Groundwater interaction Figure 3.2.: Measured and simulated concentrations in 6:2 FTSA (grey), 6:2 FTAB (green) and 6:2 FTSAam (yellow) over time in pore solutions at five depth (0.03 m, 0.5 m, 1 m, 1.5 m, 2.0 m and 2.5 m depth) in the unsaturated and saturated zones below the infiltration square during the PMP experiment. Concentrations are expressed in mgF/L. Lines represent simulated concentrations over time while dots represent measured concentrations. D2.4 – Guidance document on fate, transport and exposure for PMT’s 54 3. Soil - Groundwater interaction some hydrogeological context, this step could be complex as boundary conditions need to be reconstructed on several decades and data availability on such time length is rare. • The decrease of the pore volume in the unsaturated and saturated zones below the infiltration square This phenomenon is induced by foam formation in the pore network during AFFF infiltration. Indeed, the calibrated values of the pore volume for the cells of the unsaturated and saturated zones below the infiltration square are lower than the values estimated in the uncontaminated geomedia (Annex B.3). This decrease of the pore volume is interpreted because of the foam formation during the AFFF infiltration stage, which blocks some pore throats in the porous network. This foam seems stable during the experiment as the same values of pore volume are used to simulate solute transport during the PFAS leaching stage. Furthermore, the PFAS sorbed onto the SWI has decreased the interfacial tension, promoting water flow in pore network. These changes in hydraulic and solute transport properties of the geomedia below the infiltration square induce preferential water flow and mass transfer, explaining the quick increase and decrease of the non-reactive tracer and PFAS concentration at the beginning of the AFFF infiltration and PFAS leaching stages, respectively. Changes in hydraulic and non-reactive solute transport properties occurs only below the infiltration square and does not extend downstream. The calibrated values for the hydraulic and solute transport properties for the cells located downstream to the infiltration square in the saturated zone are similar to the ones estimated for the uncontaminated geomedia (Annex B.4). The calibration approach by performing iterative steps to calibrate first the non-reactive tracer and then the PFAS concentrations allows to identify correctly the main characteristic of the sorption processes onto the SWI and AWI in the fate and transport of the three main PFAS present in the infiltrated AFFF. Four hypotheses of the conceptual model concerning sorption reaction have been validated. • PFAS sorption reactions onto the SWI and AWI in the unsaturated zone as non-equilibrium should be used to simulate accurately the spatial and temporal changes in PFAS concentrations in this zone. The sorption reactions were non-ideal and non-linear as depicted using the Langmuir isotherm and a first-order rate equation. In consequence, the values of the model parameter cannot be compared with the partition coefficient values issued from existing literature or in silico approach as most of them account for a linear sorption reaction. • As illustrated by the higher values of the Langmuir partition coefficient and rate of the kinetically controlled sorption reaction in the upper part of the unsaturated zone compared to the deeper part, PFAS sorption onto the AWI has a key role for the three main PFAS present in the AFFF. The intensity of this sorption reaction is higher when the water content is low as the surface of the AWI increases. The model outputs (data not shown) indicate that the three PFAS are mainly sorbed in the first 30 centimetres below the infiltration square, explaining the strong decrease of PFAS concentrations in pore solution compared to the infiltrated AFFF solution. • In the saturated zone, PFAS sorption reactions are non-linear for the three main PFAS present in the infiltrated AFFF as Langmuir isotherm needs to be used to simulate D2.4 – Guidance document on fate, transport and exposure for PMT’s 55 3. Soil - Groundwater interaction Figure 3.3.: Measured and simulated concentrations in 6:2 FTSA (grey), 6:2 FTAB (green) and 6:2 FTSAam (yellow) over time in pore solutions at three depths (1.5 m, 2.0 m and 2.5 m depth) in the saturated zone in two locations (1.2 m and 2.4 m far from the centre of the infiltration square) downstream to the infiltration square in the main flow path during the PMP experiment. Concentrations are expressed in mgF/L. Lines represent simulated concentrations over time while dots represent measured concentrations. D2.4 – Guidance document on fate, transport and exposure for PMT’s 56 3. Soil - Groundwater interaction accurately spatial and temporal changes in the dissolved 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam concentrations. • The magnitude of the sorption reactions onto the SWI and AWI vary according to the PFAS present in the infiltrated AFFF. 6:2 FTSAam is the most-sorbed PFAS, followed by 6:2 FTAB. By comparison, 6:2 FTSA is less sorbed onto the SWI and AWI, explaining that this compound is the only one migrating downstream to the infiltration square. These results confirm that the molecular structure of the PFAS, including carbon chain length, hydrophobicity and nature of the functional group, are key data to understanding PFAS migration. The conclusions from the modelling of the CS#6 can be used to highlight the key processes that need to be investigated in real AFFF-contaminated sites and demonstrate that the model train can be used to build models for real AFFF-contaminated sites dealing with natural complexity and heterogeneity in terms of water flow and reactive solute transport. Key messages of the PluriMetric Pilot experiment The model train developed in the PROMISCES project can be used to simulate PFAS fate and transport in the SGW continuum in a semi-real scale experiment in controlled conditions. The selection of the key processes, that are considered in the model, is a complex task as a broad range of physical and chemical processes, and their spatial and temporal pattern in the simulated domain, had an impact on PFAS fate and transport, hence the building of the conceptual model based on available data is a key step. The use of time series measurements of PFAS to calibrate the model increases drastically its robustness. High frequency monitoring of PFAS fate and transport in the unsaturated zone is required as a sharp migration front can occur in this zone. Interrelations between physical and chemical processes controlling fate and transport of PFAS should be considered in the model as illustrated by the changes in hydraulic and solute transport properties that are impacted by AFFF infiltration. IN the PMP experiment, the AFFF was locally present in the pores as a separate phase. Water flow rate in the unsaturated zone should be carefully simulated to predict accurately migration front of the more mobile PFAS. The sorption reactions of the monitored PFAS (6:2 FTSA, 6:2 FTAB and 6:2 FTSAam) onto the soil-water interface and the air-water interface are non-linear and non-ideal in the unsaturated zone. D2.4 – Guidance document on fate, transport and exposure for PMT’s 57 4. Catchment – river – riverbank filtration – drinking water Key messages of the chapter The combination of stand-alone models in model trains expands the scope that can be covered in the context of catchment – river - riverbank filtration - drinking water interaction. Model trains can combine stand-alone models either in a complementary way or in a sequence. A complementary combination may either compare models of different complexity to find out which level of complexity (and associated efforts) is needed to answer which questions or may compare different models with their different strengths and weaknesses in parallel to assess uncertainties, and/or use models for scenario evaluation according to their specific capability. A sequential combination facilitates a broader application in terms of content and a combination of approaches at different spatial resolutions. Clearly defined interfaces are essential for a successful implementation. Examples of model trains for selected PFAS are presented for the catchment – river interaction in the urban context of the Berlin case and for the whole catchment – river – riverbank filtration – drinking water interaction on the scale of the upper Danube Basin. The Berlin case demonstrates the application of the sequential model train by combining an emission model of the city with a fate and transport model of the city´s surface waters. The Danube case demonstrates the application of a sequential model train by combining large catchment scale emission models with different types of bank filtration fate and transport models for specific locations in the catchment. Further, it also demonstrates the complementary application by comparing emission models with different strengths and weaknesses for the assessment of multiple scenarios on the catchment scale and different levels of complexity for fate and transport modelling of bank filtration. D2.4 – Guidance document on fate, transport and exposure for PMT’s 64 4. Catchment – river – riverbank filtration – drinking water 4.1. General aspects In addition to water abstraction from groundwater via springs and wells or directly from surface waters, abstraction via riverbank filtration is a commonly used method to gain raw water for drinking water supply (see Figure 4.1). Riverbank filtration is a type of filtration that works by passing water to be purified for use as drinking water through the banks of a river or lake. It is then drawn off by extraction wells some distance away from the water body. Three filtration mechanisms are possible. Physical filtration or straining takes place when colloids and suspended particulates are too large to pass through interstitial spaces between alluvial soil particles. The second mechanism is biological degradation occurring when soil microorganisms remove, and digest dissolved or suspended organic material and nutrients. The last is chemical filtration including ion exchange that may take place when dissolved chemicals adsorb to alluvial soils when passing through with water movement. Many contaminants (microbial organisms and inorganic or organic pollutants) will be removed by bank filtration, either because they get filtered out by the solid phase of the bank, because the residence time (which may be days or potentially weeks) is sufficient to render them inactive or if they are adsorbed by the soil material. Therefore, abstraction via bank filtration is a preferred option of water supply over direct use of surface water in more downstream areas of larger rivers, in cases where the catchment of the water body cannot be sufficiently protected by protection zones and bank filtration offers a natural first purification step. For instance, 45%, 50% and 16% of the total drinking water supply are provided by bank filtration in Hungary, Slovakia and Germany, respectively (Handl et al. 2017). Potential problems for water supply via bank filtration arise from the fact that filtration capacity of the riverbank is limited and depends on its width, the property of the soil as well as the physical-chemical properties of the pollutants. As water from a river (or a lake) is the source of riverbank abstraction, the whole catchment may contribute to impacts on drinking water quality, especially if persistent and mobile chemicals are considered. All different types of human activities in the catchment are potentially acting as sources or pathways of contamination of the river and connected water supply via bank filtration (industrial and agricultural activities, chemical use in households, waste and wastewater disposal...). Regarding circular economy routes, drinking water supply via bank filtration is considered as “semi closed water cycle” as industrial and urban waste waters are discharge into the rivers and indirectly and unintended reused for drinking water supply after mixing and dilution in the natural runoff of a region. Figure 11 shows the schema of circular economy route “semi-closed water cycle” in the context of catchment – river – riverbank filtration – drinking water interaction. Exposure assessment of drinking water from bank filtration via emission, fate and transport modelling requires a model train covering the whole catchment – river – riverbank filtration – drinking water interaction. That means it should on the one hand consider the sources and pathways of pollutants in the catchment, the transport and fate of the chemicals in the river system, as well as the behaviour and fate during the filtration step in the riverbank of the river to the water abstraction. This offers the chance to explain reasons for pollution levels in drinking water and to calculate scenarios for potential future developments as implementation of measures, changes in chemical use, accidental spills or climate change. In the upcoming chapters we focus on emission models at catchment scale, on emission modelling in an urban D2.4 – Guidance document on fate, transport and exposure for PMT’s 65 4. Catchment – river – riverbank filtration – drinking water context, on fate and transport modelling at bank filtration sites and present application cases for Berlin and the upper Danube catchment. Figure 4.1.: Schema of the circular economy route “semi-closed water cycle” in the context of catchment – river – riverbank filtration – drinking water interaction. 4.2. Emission models on catchment scale in the context of model trains 4.2.1. General Emission models at the watershed scale are typically used by river basin management (RBM) authorities as part of the implementation of the Water Framework Directive (WFD), the main EU regulatory instrument for water management (2000/60/EC). Emission modelling can be used for Priority Substances (PS) and Priority Hazardous Substances (PHS), for which environmental quality standards (EQS) have been defined (EC, 2013) at the European level. Compliance with these EQS is required to achieve “good chemical status”. Emission modelling can also be used for pollutants of regional or local relevance (River Basin Specific Pollutants, RBSP), for which EQS need to be established and compliance achieved as part of achieving “good ecological status”. Finally, emission modelling can address various Contaminants of Emerging Concern (CEC), which are not (yet) regulated. The WFD requires the preparation of an inventory of emissions, discharges and losses of all PS and PHS that are found relevant. Such D2.4 – Guidance document on fate, transport and exposure for PMT’s 66 4. Catchment – river – riverbank filtration – drinking water inventories should inform RBM authorities on the loads discharged to the aquatic environment. In 2012, Guidance Document No. 28 on the Preparation of an Inventory of Emissions, Discharges and Losses of Priority and Priority Hazardous Substances was issued, as part of the Common Implementation Strategy for the European Comission (EC) 2012. This introduction is based on the information compiled in this Guidance. An emission inventory should be seen as a tool, which may be used to: • provide quantitative understanding of present emissions (diagnosis), which is a prerequisite for defining (cost-)effective control measures; • appreciate knowledge and data gaps and subsequently provide focus for future research and data collection; • conduct scenario analyses (prognosis) to assess the efficacy of pollution control measures and/or the impact of external changes (e.g. climate change & adaptation measures). Especially for the latter objective, emission modelling is combined with water quality (fate & transport) modelling, to translate the emission inventory into in-stream concentrations. For the elaboration of the emission inventory, the Guidance discusses four so-called “tiers”: 1. Tier 1 comprises an inventory of point sources using statistical data including point source information reported under the European Pollution Release and Transfer Register (E-PRTR). 2. Tier 2 adds a quantification of in-stream loads, based on concentration and discharge data. The resulting riverine load, in combination with the information gained in tier 1 allows the allocation of observed loads to point and diffuse sources. A high contribution of diffuse sources is a reason to proceed to tier 3 or 4. 3. Tier 3 is a pathway-oriented modelling approach, using more specific information about land use, hydrology and basic transport processes of pollutants towards the surface water. Regionalized emissions for small catchments (analytical units) are calculated and subsequently aggregated to larger units. 4. Tier 4 is the source-oriented modelling approach, that sets up a mass balance or Substance Flow Analysis (SFA) and addresses the whole system starting from the principal sources of substance release. Figure 4.2 provides a schematic overview of Tier 2, 3 and 4 and their inter-relation. D2.4 – Guidance document on fate, transport and exposure for PMT’s 67 4. Catchment – river – riverbank filtration – drinking water Pathways P1 Atmospheric deposition directly to surface water P7 Storm water outlets and combines sewer overflows + unconnected sewers P2 Erosion P8 Urban wastewater treated P3 Surface runoff from unsealed areas P9 Individual – treated and untreated – household discharges P4 Interflow, drainage and groundwater P10 Industrial wastewater treated P5 Direct discharges and drifting P11 Direct discharges from aquaculture, fisheries and other instream activities P6 Surface runoff from sealed areas P12 Natural background Figure 4.2.: Overview of riverine load (Tier 2), pathway-oriented (Tier 3) and source-oriented (Tier 4) approaches. Modified from Chorus and Zessner (2021) and European Comission (EC) (2012). D2.4 – Guidance document on fate, transport and exposure for PMT’s 68 4. Catchment – river – riverbank filtration – drinking water 4.2.2. Modelling approaches The PROMISCES project applied models in support to both Tier 3 and Tier 4 approaches, covering the Danube River Basin upstream of Budapest. A compact comparative assessment of the two approaches is shown in Table 4.1. Table 4.1.: Compact comparative assessment of MoRE model and PPM model applied in the upper Danube Catchment. Modelling of Regionalized Emissions (MoRE) PROMISCES watershed model for PM substances (PPM) Pathway-oriented (Tier 3; WFD Guidance) Source-oriented (Tier 4; WFD Guidance) 526 analytical units with an average size of 350 km2and a single (in specific cases two) outflow point and an upstream downstream connection Regular grid consisting of 81,217 cells (∼1500 ×1500 m) Simulates annually/seasonally steady states Simulates day-to-day variability Operates on annually averaged hydrology information (generated by the Tier 4 model) Calculates the time dependent hydrology using global and local weather data Calculates average emission loads via different pathways (see Figure 4.2). For PFAS diffuse emissions from legacy pollution (PFAS production and inhabitant related pollution as firefighting grounds eg. at airports and landfills) have been implemented for the first time. Calculates daily time series of emissions Calculates annual average water quality with potentially using retention parameterized on hydrological indicators Calculates water quality using a dynamic fate & transport model Specific input data used: substance concentrations in simulated pathways Specific input data used: substance flux (mass/time) released from simulated sources (see Figure 4.2) D2.4 – Guidance document on fate, transport and exposure for PMT’s 69 4. Catchment – river – riverbank filtration – drinking water MoRE PPM Model is validated on in-stream loads and concentrations Model is validated on in-stream concentrations Scenario assessment of the model mainly addresses measures focusing on specific pathways. In combination with climate driven water balance models, climate scenarios can be implemented The model is capable of assessing scenarios focusing on reduction of emissions via source abatement, may calculate effects of accidental spills and implement climate scenarios. Runs on Windows-based computers. Offers graphical user interface, but still technically challenging. Runs on Windows-based personal computers. No graphical user interface available for less-experienced users. Technically more challenging. Both models use the following spatial information and data (though they often express it differently): (a) catchment delineation and digital terrain model; (b) land use, population and settlement types; (c) soil loss; (d) water consumption; and (e) stormwater and wastewater collection and treatment infrastructure. Both models apply the principle of mass balances on (sub) catchment scale, which calculates the instream loads of the river net (I) along the flow tree based on the sum of emission loads (E) minus substance degradation (D) and retention (R) and the substance concentrations (c) based on instream loads (I) at a specific point subdivided by the river flow (Q): I= ΣE−D−R(4.1) c=I/Q (4.2) Conceptual differences result in the fact that both models have complementary strengths and weaknesses. In respect to application by river basin management (RBM) authorities MoRE (Fuchs et al. 2017) is characterized by much stricter implementation and documentation requirements reducing the flexibility but making used input information, calculation steps and results more transparent, while the PPM model (Verseveld et al. 2024; DELTARES 2025; Smits and Beek 2013) as expert tool has its strength in a higher flexibility during application which reduces the transparency. In respect to resolution the focus of the PPM model clearly lies on the time resolution as it is driven by daily time steps of a hydrological model. The focus of MoRE model is much stronger on substance specific data of different pathways and therefore on the regional representation of these pathways on the level of analytical units. The complementary foci are of clear advantage for a combined application in a model train. While MoRE highly benefits from the PPM model by high resolution water balance modelling and resulting flow data including the possibility of climate driven runoff scenarios, the PPM model benefits from data on emission pathways from MoRE either for validation of its calculations of pathway emissions or for back-calculating emission factors of emission sources and is able to disaggregate this information to calculations of river concentrations with daily time step. D2.4 – Guidance document on fate, transport and exposure for PMT’s 70 4. Catchment – river – riverbank filtration – drinking water The latter is a requirement in case scenarios include more dynamic changes with time as it is the case in the accidental spill scenario presented in section 4.6.4. More details for the MoRE application in the context of this guidance document can be found in Annex D and for the PPM model in Annex E. 4.3. Emission model in an urban context 4.3.1. General aspects Precipitation events can introduce large quantities of various pollutants, nutrients and pathogens into surface waters. Urban water bodies with densely populated and highly sealed catchments are often particularly affected by stormwater discharges from the separate sewer system and combined sewer overflows (CSOs). Quantitative estimates of the chemical loads in stormwater and wastewater runoff are an important basis for the assessment of water pollution. This chapter presents a method to assess the rain-related PMT/PFAS pollution situation of surface waters caused by urban drainage inputs via source-pathway and process descriptions. Based on measured data and/or additional literature data with a monthly resolution, monthly loads of substances entering surface waters via the three pathways of stormwater discharges from separate sewer systems, wastewater treatment plants and combined sewer overflows are calculated (Figure 4.3). For this purpose, an emission model has been developed specifically for an urban catchment to calculate PFAS and PMT loads. The calculated loads are then used in a surface water model to predict the fate and transport of these substances in the surface water systems. As the substances may adsorb to particulate organic material, it is possible that in areas (or periods) with very low flow velocities, sedimentation of the organic material and the adsorbed pollutants occurs. Adsorption will likely not affect all PMTs and certainly not in the same amount, so that some variation has to be considered. 4.3.2. Emission model A number of input parameters are required for the emission model to calculate the monthly load of pollutants entering various water bodies and watercourses via stormwater and wastewater via the separate sewer system, combined sewer overflows (CSOs) and wastewater treatment plant (WWTP) effluents: 1. Mean concentrations in stormwater runoff 2. Mean concentrations in wastewater (+ CSO) 3. Mean concentrations in WWTP effluent 4. Stormwater runoff volumes (separate sewer system), in m3/month for defined water bodies 5. Wastewater volumes (WWTP effluent & CSO* volumes) discharged to receiving water bodies, for defined water bodies *CSO volumes are estimated at 0.3% of wastewater volumes D2.4 – Guidance document on fate, transport and exposure for PMT’s 71 4. Catchment – river – riverbank filtration – drinking water Figure 4.3.: Considered substance sources and input paths of watercourse pollution Following the calculation of the monthly loads, the outputs of the model are total emission loads to each water body subdivided into different categories, depending on the source of pollution (stormwater or wastewater) and/or its pathway (separate sewer system, CSO or WWTP) with a monthly resolution. Thus, for each water body, the total monthly load of a substance in the stormwater runoff and in the wastewater is determined, as well as the proportion of these loads accounted for by combined sewer overflows, stormwater and treated wastewater discharges. 4.3.3. Surface water model Following input is required for the subsequent surface water model: 1. The network of channels, rivers and lakes that form the surface water system of Berlin as an example for model application. This network is schematised so that it is useable by the water quality model. 2. The hydraulic information that makes up the flow regime. a) Inflow from the rivers (in this case notably Spree and Havel) and the distribution of the water via the channels; b) Groundwater abstraction for drinking water production, known per calendar month, in m3/month; c) Bank filtrations shares, again known or estimated per calendar month, in m3/month; D2.4 – Guidance document on fate, transport and exposure for PMT’s 72 4. Catchment – river – riverbank filtration – drinking water d) Stormwater runoff volumes (separate sewer system), in m3/month for defined water bodies e) Wastewater volumes (WWTP effluent & CSO*, receiving water bodies) 3. Concentrations of the various substances in the wastewater effluent. Averages are deemed to be representative of typical circumstances. Nevertheless, if the storm events show significant contributions (in terms of the mass that enters the system during the event versus the average/typical mass in the same duration), a detailed time series which resolves the storm events and the associated pollutant concentrations might be a more appropriate approach. This presumes that data are available that allow us to draw such a conclusion. For the PMT substances we need to consider different adsorption characteristics, as they exhibit a wide range of behaviours (Oudega et al. 2024). Adsorption of the PMT substances to organic or inorganic particulate matter means that under low flow-velocity circumstances, the adsorbed fraction can sediment onto the bottom of the water body along with the particulate matter. This could lead to a build-up of these substances in the sediment. By using various such characteristics, we get better insight in the role of sedimentation with regards to the transport and fate of PMT substances. The expected output of the surface water model can be summarized as follows: 1. Concentration patterns as they evolve in the water system in place and time. 2. Indications of where the PMT substances end up in the sediment and how much. 3. Mass balances indicating the importance of storm water versus wastewater and background concentrations. 4.4. Fate and transport modelling at bank filtration sites 4.4.1. Basic ideas The bank filtration (BF) models described in this chapter were devised to answer questions regarding PMT (here we focus on PFAS) transport and fate in riverbank filtrations systems (RBF). The focus herein lies in predicting PMT concentrations in drinking water wells in RBF systems. This is done by simulating infiltration of PMT into the riverbank and subsequent transport through the groundwater body towards the pumping well. By doing this, a variety of questions might be answered: •Future changes: How would PMT concentrations in a drinking water well change if . . . ... the pumping rate changes (population growth)? ... PMT production increases (worst-case scenario)? * ... certain legal requirements come into play (e.g. further treatment of wastewater)? * ... other local changes are carried out (e.g. river water level changes)? D2.4 – Guidance document on fate, transport and exposure for PMT’s 73 4. Catchment – river – riverbank filtration – drinking water monthly loads were then calculated for a range of 1000 random concentrations and volumes, and the results were summarised in statistical key figures. As an exemplary result, Figure 4.5 shows the annual discharge volumes into Berlin's watercourses by source (rainwater and wastewater) and path (separate sewer system, CSO, WWTP) next to calculated monthly loads of PFOA. In terms of both volume and load, WWTP treated wastewater is the largest source. Figure 4.5.: Left side: Monthly PFOA loads by path into Berlin’s surface waters, right side: Contributions of the separate sewer system (Sep. sewer), wastewater treatment plants (WWTP) and combined sewer overflows (CSO) to annual rainwater and wastewater discharges. 4.5.3. Explanation of the surface water model Based on geometrical information of the components of the water system (length and width of branches, depth, size of the various lakes and the connections between the components), a somewhat simplified schematization was devised (see Figure 4.6). The flow through the rivers and canals was estimated using available hydraulic data (Schumacher 2023). An important aspect of the flow field is that it must be volume-conserving: if two branches join up together, then the flow in the downstream branch should be the sum of the flows in the two upstream branches. When a branch splits up, a similar condition holds. To determine a consistent, volume-conserving flow field with the appropriate geometrical information and connections, an auxiliary program was used. Each branch consists of two or more grid cells, depending on the length of the branch. The various lakes were treated as if they were wide branches. The exception to this set-up is the Flughafensee. As it is an isolated lake with a depth of up to 36 m, an alternative approach was used: it is assumed that the lake is well-mixed in the horizontal but exhibits concentration gradients in the vertical. While there are no detailed measurements regarding temperature available, it seems likely that the lake is not well-mixed in the vertical direction, see Figure 4.7. D2.4 – Guidance document on fate, transport and exposure for PMT’s 80 4. Catchment – river – riverbank filtration – drinking water Figure 4.6.: Layout of the Berlin surface water model, visualized via the concentration of a tracer. (Note: not all connections are visible, notably the connection between the Wannsee and the Unterhavel, the lower part of the Havel.) D2.4 – Guidance document on fate, transport and exposure for PMT’s 81 4. Catchment – river – riverbank filtration – drinking water To illustrate the model the input of suspended sediment was kept constant in time and the concentrations resulting after a long time were plotted. The sediment will actually accumulate on the bottom, so there is no equilibrium there, but in the water column a steady distribution is reached. Figure 4.7.: Schematic representation of the Flughafensee as a column model. The quantity shown is the concentration of suspended solids (mg/L). The concentration at the surface is lower than near the bottom due to sedimentation. 4.5.4. Modelling the PMT substances The approach to modelling the PMT substances in water and sediment is the same for both schematics: PMT behave as dissolved substances but can also adsorb to organic and to a lesser extent inorganic particulate matter. The following assumptions are made: • PMT substances do not degrade and are not volatile, so that they persist in the water system. • Adsorption to particulate matter can be described via a partition coefficient, so that one fraction is dissolved, and the other is adsorbed. The adsorption process is assumed to be fast, so that an equilibrium exists. • The adsorbed fraction can sink to the bottom along with the particulate matter and accumulate there. Thus, we need to include the following substances: •Particulate matter in the water phase and at the bottom D2.4 – Guidance document on fate, transport and exposure for PMT’s 82 4. Catchment – river – riverbank filtration – drinking water •PMT concentration in the water phase and in the bottom sediments The model for the Berlin surface waters was used to make a first, illustrative calculation for PFOA (see Figure 4.8). The results of the month July were used, assuming a constant influx. As the residence time of the system is rather long (several months, partly due to the lakes that are an integral part of the water system), the simulation time was set to 5 months. Figure 4.8.: Simulated concentrations of PFOA in Berlin watercourses as the result of rainwater, treated wastewater and raw wastewater discharges (in ng/L). The calculation starts out with a concentration of zero and the inflow of the various rivers (Havel, Spree and others) into the system, as well as the discharges of (treated) wastewater at the various branches, cause the pollutant to enter the surface water. Due to the adsorption to particulate matter (organic and inorganic) of PFOA, part of it can reach the sediment layer. In the Berlin water system, the flow velocity is very low, which makes it likely that PFOA in the sediment continually accumulates over time. Figure 4.8 shows the simulated PFOA concentrations in ng/L in Berlin’s watercourses. Higher concentrations can be observed in the Rudower-Teltowkanal, which receives the effluent from two large sewage treatment plants. The high concentrations are partly due to the high discharge volumes of the WWTPs and partly to the low flow velocities in the Teltowkanal. The various treatment plants, for instance in the Teltowkanal, cause an increase in the downstream concentration. The magnitude of this increase depends on the flow rate and on the actual discharge. D2.4 – Guidance document on fate, transport and exposure for PMT’s 83 4. Catchment – river – riverbank filtration – drinking water Key messages of the Berlin Case Study The largest emission loads of most PFAS and PMT substances analysed enter the water bodies via the treated wastewater discharges of the WWTPs. While the concentrations of PFOA are in the order of 0.02 ng/L, this concerns a single PFAS species only. The total PFAS concentration is likely to be an order of magnitude larger. Water systems like that of the city of Berlin are quite complex and it is inevitable that some simplifications are made when schematising them. Moreover, important data are often missing or are only approximately known, such as the typical concentration of particulate matter. One has to make informed assumptions, then, even if it introduces further uncertainties. 4.6. Model train application in a large catchment 4.6.1. Description of the study site The study site for application of a model train for exposure assessment in the context of the catchment – river – riverbank filtration – drinking water interaction in a large catchment is the Danube catchment upstream of Budapest shown in Figure 4.9. Some basic data for characterization of the catchment are shown in Table 4.3. The main questions addressed in this study site are (i) what are the main sources and pathways of PFAS pollution in the catchment, (ii) how do different PFAS parameters behave during bank filtration, (iii) how could future changes in the catchment lead to changes in concentrations in pumping wells for drinking water abstraction, and (iv) is there a risk of failing to achieve standards for drinking water or (E)QS in surface or ground water now or in the future? The model train applied consists of the combination of the MoRE and PPM models for assessing sources and pathways of PFAS emissions in the catchment as well as 3D physically distributed reactive flow and transport models (MODFLOW/MT3DMS) for three bank filtration sites, namely one small-scale site (~10 m) in Vienna and two larger-scale (~100-2000 m) sites on Szentendre Island in Budapest. The catchment models focus on the PFAS parameters: PFBA, PFPeA, PFHxA, PFHpA, PFOA, PFBS, PFHxS, PFOS, ADONA and HFPO-DA/GenX. These ten parameters are fully reflected in catchment modelling with MoRE and PPM model, with the possibility of validation against measurements. The models can calculate emission loads for PFNA, PFDA, PFPeS, PFNS, PFDS, 6:2 FTS PFOSA, N-EtFOSAA as well, due to quantifiable concentrations in some pathways. Nevertheless, concentrations of these parameters in rivers are mostly below limit of quantification. Therefore, modelled river loads cannot be validated by comparison to observations. The behaviour of the first ten substances is further investigated using bank filtration modelling. While in Vienna most of drinking water is supplied from karstic springs, in Budapest Danube bank filtration is the dominant drinking water source. The bank filtration at Danube Island is a potential source to increase the water supply as a rising demand is predicted for the future. Water is abstracted from multiple wells groups on both the eastern and western bank D2.4 – Guidance document on fate, transport and exposure for PMT’s 84 4. Catchment – river – riverbank filtration – drinking water Figure 4.9.: Map of the Danube catchment upstream of Budapest, showing monitoring points for validation at Danube and its tributaries as well as bank filtration sites at Vienna and Budapest. Table 4.3.: Characterization of the catchment upstream bank filtration sites in Vienna and Budapest Danube Island Vienna Danube Island upstream Budapest Catchment km2101,527 183,293 Inhabitants Inh. 15,577,100 23,858,775 Waste water from municipal and industrial WWTPs p.e. 18,482,423 30,911,725 Waste water from municipal and industrial WWTPs 106m3/a 1,349 2,256 Average river discharge m3/s 1,871 2,333 Average river discharge 106m3/a 59,004 73,573 D2.4 – Guidance document on fate, transport and exposure for PMT’s 85 4. Catchment – river – riverbank filtration – drinking water of Szentendre Island. One abstraction well from the well group “Tahi” and one from the well group “Surány” have been chosen for detailed evaluation due to favourable conditions with respect to accompanying monitoring wells. Figure 4.10 shows a map of the local situation. Figure 4.10.: Map of bank filtration sites at Budapest. Left: ground level [m.a.s.l.], Right: detailed map of Site 1 (Tahi I-5) and Site 2 (Surány 12). Extraction wells in red, monitoring wells (MW) in orange, unused monitoring wells in green. For a smaller scale BF study, the Danube island in Vienna was chosen (Figure 4.11). This is an area close to where bank filtered water is used to produce drinking water. Three monitoring wells (MW) are located in a straight line at distances of 1, 15 and 28 m from the riverbank. Besides the smaller scale, another difference with the Budapest transects is the absence of a pumping well (PW) and the presence of a significantly clogged riverbank, potentially increasing the effectiveness of BF at the site. D2.4 – Guidance document on fate, transport and exposure for PMT’s 86 4. Catchment – river – riverbank filtration – drinking water Figure 4.11.: Map of bank filtration site at Vienna. Left: model boundaries and absolute elevation (m.a.s.l.) of the Danube Island in the center of the city, Right: detailed map of the transect. Groundwater Wells (GW) 1, 3 and 4 were monitored. No extraction wells are present in the transect. 4.6.2. Catchment emission models In the MoRE application for the upper Danube catchment, emission loads via different pathways are calculated at the level of 526 sub-catchments as analytical units with an average size of 350 km 2 and summed up along the flow tree to assess river loads at the outlet at each subcatchment. If river concentrations are observed by monitoring, modelled concentrations can be compared to these observations to validate model calculation. In the upper Danube Basin, river concentrations have been observed at 11 locations (see Figure 4.9) and river loads have been calculated from that and associated flow data. Figure 4.12 shows model performance for the sum of 10 parameters covered by quantitative modelling (see above) calculated according to EU Drinking Water Directive (EU DWD) and the suggested procedure according to the proposed new groundwater and environmental quality standard directive (EC COM(2022) 540 final). Model efficiency coefficients (KGE and NSE) > 0.65 indicate high model performance, with a good agreement between modelled and observed loads. Due to the good model performance, it is justified to use model results for more detailed evaluation and scenario calculation. D2.4 – Guidance document on fate, transport and exposure for PMT’s 87 4. Catchment – river – riverbank filtration – drinking water Figure 4.12.: Model validation for the MoRE model at outlets of 11 sub-catchments for different PFAS. X and Y axes in logarithmic scale. KGE and NSE indicate the model performance. Values > 0.65 indicating a good agreement between modelled and observed loads The dominating emission pathways of the catchments upstream of each of the eleven validation points are shown in Figure 4.13. There is a clear dominance of industrial direct discharges into the rivers of the Danube catchment upstream of Budapest for most carbonic acids (PFBA, PFPeA, PFHxA and PFHpA as well as ADONA and GenX). These emissions are mostly associated to the emissions from 3M PFAS production in the Chemical-Park in Gendorf close to the Alz river. PFOA production at this location was stopped 15 years ago. Thus, PFOA point source emissions from this place are currently neglectable. Nevertheless, due to contamination of the surroundings, diffuse legacy pollution from this area is still the dominating pathway for PFOA. Another emission pathway with a potentially high share of emissions is the legacy pollution from urban areas (P4c: inhabitant-specific diffuse pollution via groundwater), which includes pollution from firefighting training grounds at airports or other locations as well as from former landfills. The estimates for this pathway are however still affected by high uncertainty due to lack of specific input data. The contribution of municipal wastewater treatment plant effluents is between 10 to 25% of the total emission of different PFAS substances. For some of the parameters also background loads via groundwater contribute with relevant shares. The overall assessment of emission pathways highlights the relevance of 3M PFAS production at the Chemical-park Gendorf as well as different types of legacy pollution for the overall loading of river Danube at the bank filtration sites at Vienna and Budapest. D2.4 – Guidance document on fate, transport and exposure for PMT’s 88 4. Catchment – river – riverbank filtration – drinking water Figure 4.13.: Modelled contribution of different emission pathways to loads of 10 single PFAS at outlets of 11 sub-catchments for different PFAS. Black dots indicate the river load calculated from observed concentrations and river discharge. Using the validated models for the upper Danube Basin, a risk map has been drawn showing the risk of exceedance of thresholds from EU legislation. Figure 4.14 shows a base variant of the model results. The worst-case and best-case variants of these calculations, which account for the uncertainties in emission modelling, can be found in Annex D. D2.4 – Guidance document on fate, transport and exposure for PMT’s 89 4. Catchment – river – riverbank filtration – drinking water degradation, transformation), a concentration of 0 was assumed at the model boundaries for all 10 species. Behind the PW at both transects though, some substances were found to have locally elevated concentrations (especially PFOA, PFOS and Gen-X), presumably due to an unknown inland source. To accommodate for this, a daily concentration input was applied behind Surány MW4 and Tahi MW3. Additionally, because the 10 species of PFAS have been present in the environment for much longer than the calibration period, a “warm-up” period of at least 1 year was run for each scenario to obtain the initial spatial concentration distribution of each of the 10 PFAS. At Tahi, the simulated median PFAS concentrations for the monitoring period show a good fit compared to the observed values (Figure 4.16). Most PFAS agree within ±1 ng/L. Some substances (e.g. PFPeA, PFHxA and ADONA at Tahi) show too high modelled median groundwater concentrations, most likely due to the input concentrations in the Danube not fitting exactly to the measured values (Groot et al. 2025). A lack of variation can be seen when comparing observed to simulated concentrations. Likely, this is an effect of lack of temporal correlation in the randomly generated daily river input concentrations. Because daily river concentrations can have any value within the chosen statistical distribution, concentrations jump up and down from day to day. These concentrations mix in the groundwater, so that variability is reduced. In other words, there are no longer periods of low or high concentrations in the model, that most likely would have occurred in reality. However, due to monthly sampling, we have no information on this. Lastly, elevated background concentrations (e.g. of PFOA, PFOS, GenX in Surány) are not accurately simulated because of the little information available on these inland sources. After calibration, the model was run for the period 2015–2021, which represents the reference period for the scenario simulations. For this period, daily Danube PFAS concentrations were generated on the same statistical basis as for the monitoring period (see above, Groot et al. (2025)). Subsequently, scenarios were simulated, of which the results can be read in Section 4.6.4. Differences between the small-scale and large-scale BF modelling There are many differences worth noting between the Vienna (small-scale, ~ 10 m) and Budapest (large-scale, ~ 100–1000 m) sites. In Budapest, because of the pumping of drinking water wells, the sampled groundwater is a mix between water that has infiltrated a long time ago (possibly even from the other side of the Szentendre island) and water that has infiltrated recently from the nearby riverbank. In Vienna, the water infiltrates only from one side of the island and travels a shorter distance to the wells, leading to less mixing. Therefore, the groundwater represents more directly the PFAS concentrations in the Danube. This can be easily seen from Figure 26, where concentrations are more stable than in the Budapest sites. The shorter travel distances and simpler, unidirectional flow system make the Vienna case better defined with smaller uncertainties. This is reflected in the better calibration result (i.e. residuals between measured and modelled heads), as was shown in Table 4.5. Another difference is the presence of a clogging layer in Vienna, which potentially increases effectiveness of the RBF system. However, as can be seen from Figure 4.16, PFAS concentrations do not significantly decrease from the river to the groundwater. This is because of the continuous influx of PFAS from the river, meaning that over time, an equilibrium developed between the river water and groundwater concentrations. Because of this continuous influx, the clogging D2.4 – Guidance document on fate, transport and exposure for PMT’s 96 4. Catchment – river – riverbank filtration – drinking water Figure 4.16.: Box plots of measured versus simulated concentrations of all 10 substances in monitoring wells (MWs) and the pumping well (PW) from 2022–2024 at Vienna (top), Surány (middle), and Tahi (bottom). The black line indicates the median, the boxes the quartiles and the whiskers represent the smallest and largest values within 1.5 times the inter-quartile range from the first and third quartiles. Dots represent outliers. D2.4 – Guidance document on fate, transport and exposure for PMT’s 97 4. Catchment – river – riverbank filtration – drinking water layer does not provide an extra barrier against PFAS contamination. However, in the case of a temporary influx (e.g. in the case of an accidental spill of a large amount of PFAS), it might prove valuable. More on this can be read in section 4.6.4. Despite these systemic differences between these sites, in terms of modelling approach they can be treated similarly and the chosen 3D numerical approach is appropriate for both. The monitoring data, which do not provide information on the reactive behaviour of PFAS, also do not justify the use of separate modelling approaches for the two scales in order to get the most information from the modelling exercise. 4.6.4. Model train application for scenario assessment Scenarios assessed with the model train are not intended to be realistic predictions of future development, as this was not feasible with the available data and models. Instead, the scenarios are intended to show how hypothetical situations would be reflected in model results, and what kind of questions could be answered by the model train if future developments were better known. Table 4.7 gives an overview on scenario assumptions of changes in the upper Danube catchment to be assessed with the model train prepared for the catchment – river – riverbank filtration – drinking water interaction. As reference situation we use the period 2015–2021, for which most recent multiyear hydrological information was available for the whole catchment. As for the main PFAS point source in the catchment, the 3M PFAS production in Gendorf, the company announced the phase out of PFAS production in 2025, the ending of the point source emission stemming from this production is implemented in a Baseline Scenario (BL) which is otherwise unchanged compared to the emissions in the reference situation. Also, the diffuse emissions from the Gendorf site stay unchanged. Further on, the BL scenario is used to show effects of contrasting climatic conditions in the catchment. It assumes hydrological conditions of the year 2013 as an example for a wet year or a “pre-climate change situation” (low flow Q 95 at Vienna: about 1500 m 3 /s, as compared to the long term average low flow Q 95 of 950 m 3 /s) and it assumes hydrological conditions of the year 2018 as an example of the dry year or a “post climate change situation” (low flow Q 95 at Vienna: about 700 m 3 /s as compared to the long term average low flow Q95 of 950 m3/s) in the catchment. The scenario “accidental spill” (AC) assumes a fire close to the city of Linz upstream of the bank filtration sites at Vienna and Budapest, leading to a spill in the river of PFAS contained in the firefighting foam. In this scenario two contrasting recipes are assumed for firefighting foams, an old one based on PFOS and a newer one base on 6:2 FTS. In both cases, “worst case” assumptions for the spill have been made to visualize what might happen in an extreme situation. The spill is tracked on its way down river Danube and its impact on bank filtered water is assessed. Again, two contrasting hydrological conditions in the catchment controlling the water flow and PFAS transport are shown as sub-scenarios. Water pollution control scenarios assume two levels of action to reduce PFAS emissions in the catchment and again two contrasting flow regimes. The first level of water pollution control, implemented in scenario “WPC1”, is addressing current environmental emissions via wastewater treatment plants (WWTP) and is assuming advanced quaternary treatment with activated carbon at all WWTPs with more than 10,000 p.e. (population equivalents). The second level of D2.4 – Guidance document on fate, transport and exposure for PMT’s 98 4. Catchment – river – riverbank filtration – drinking water Table 4.7.: Overview of scenario assumptions to calculate potential changes of PFAS concentrations in river Danube and bank filtration wells at Vienna and Budapest. Scenario group Scenario Climate/ river discharge PFAS production Gendorf Legacy pollution Accidental spill Advanced waste water treatment PFAS use restriction Reference period (Ref) based on historical data (2015– 2021) Yes current situation No No change No change Baseline (BL) BL-preclimate change hydrology 2013 No current situation No No change No change BLpostclimate change hydrology 2018 No current situation No No change No change Accidental Spill (AC) AC1high flow high-flow period starting 2013-01-09 No current situation Spill near Linz, new “6:2 FTS foam” No change No change AC1low flow low-flow period starting 2018-07-11 No current situation No change No change AC2high flow high-flow period starting 2013-01-09 No current situation Spill near Linz, old “PFOS new foam” No change No change AC2low flow low-flow period starting 2018-07-11 No current situation No change No change Water pollution control I (WPC1) WPC1pre hydrology 2013 No current situation No activated carbon at WWTP > 10k PE No change WPC1post hydrology 2018 No current situation No No change Water Pollution Control II (WPC2) WPC2pre hydrology 2013 No Groundwater remediation at hot spots No activated carbon at WWTP > 10k PE Restriction to essential use concept WPC2post hydrology 2018 No No D2.4 – Guidance document on fate, transport and exposure for PMT’s 99 4. Catchment – river – riverbank filtration – drinking water water pollution control, implemented in scenario “WPC2”, addresses the much more challenging task of remediation of legacy pollution from contaminated sites and a source control strategy based on the concept of an essential use for PFAS (PFAS may only be used in products where their application is considered as “essential”). More details on the assumptions for the different scenarios can be found in Annex G. An overview of the modelling results of scenarios is shown in Table 4.8. As in Figure 4.14, a risk-ranking schema based on defined concentrations ranges from very low to very high risk for exceedance of the QS for the PFAS sum parameter from the EU Drinking Water Directive (EU DWD) or the QS (quality standard) included in the proposal for a new groundwater directive and the EQS (environmental quality standard) included in the proposal for a new environmental quality standard directive (EC COM(2022) 540 final) have been used for assessment. Again, sums have been calculated from the 10 parameters where quantitative modelling was possible. The risk of exceedance of the quality standards of the DWD is very low in most of the scenarios. An exception is the accidental spill scenario (AC) with its sub-scenarios. Assuming a spill with the newer 6:2 FTS based firefighting foam would only cause a medium or low risk of exceedance of the sum of PFAS in river Danube for few days. The effect of bank filtration would lower this risk for drinking water abstraction by flattening the peak of the spill by retardation and mixing processes. The example of the spill with the old PFOS-based foam clearly indicates the risk that would have been provoked by such a spill. Depending on flow conditions, a high or very high risk of exceedance of drinking water standards would have been expected in river Danube at Vienna and Budapest for some days. Again, the effect of bank filtration would have reduced the risk for exceedance of quality standards depending on the flow time in the groundwater being significantly higher at water abstraction wells in Surány than in Tahi or at the control well at Vienna (the later one not used for water abstraction). The risk of exceeding the (E)QS for PFOA-eq as proposed for a new groundwater and environmental quality standard directive is much higher than in the DWD case. While the closure of the 3M Gendorf production site and the implementation of further water protection control measures (WPC1 and WPC2) would lead to a reduction of the risk of exceedance of the EQS in the Danube, this risk might increase again in case of hydrological conditions causing a reduced flow in the river. A reduction of the risk for exceedance of the proposed QS for the sum of PFOA-eq in groundwater as effect of bank filtration is insignificant for all scenarios, as this QS is calculated as average over the whole year and the retardation of bank filtration can reduce peak pollution, but does not hinder PFAS to distribute on a longer run. In contrast, inland pollution from the Danube Island at Budapest Surány may even contribute to the risk of exceedance of QS in groundwater in case of low water levels in river Danube and higher shares of contribution from inland water to the abstraction wells. D2.4 – Guidance document on fate, transport and exposure for PMT’s 100 4. Catchment – river – riverbank filtration – drinking water Table 4.8.: Estimated concentrations and risk of exceedance of (E)QS in river Danube at Vienna and Budapest and after bank filtration for different scenarios for the 10 PFAS evaluated. Maximum concentrations (ng/L) were reported for the DWD, average concentrations (ng PFOA-eq/L) for the (E)QS. ΣPFAS10 DWD (EU DWD) ΣPFAS10 PFOA-eq draft (E)QS (EC COM(2022) 540 final) Vienna Budapest Vienna Budapest Scenario Danube BF Danube Tahi Surány Danube BF Danube Tahi Surány Ref 36.8 19.9 41.0 26.6 24.6 5.2 3.2 6.3 6.3 8.5 BLpre-cc 11.5 5.4 12.4 21.3 27.1 3.0 1.6 3.8 4.6 11.2 BLpost-cc 17.7 8.4 18.8 14.3 20.9 5.2 2.9 6.4 6.4 11.6 AC1-hf for 5 days 6.5 for 5 days 18.2 for 92 days average 2.2 (>365 days) average 4.3 (205 days) 9.1 (209 days) AC1-lf for 5 days 10.0 for 5 days for 25 days for 32 days average 3.7 (>365 days) average 6.8 (>365 days) 9.8 (>365 days) AC2-hf for 3 days 8.1 for 8 days for 27 days for 144 days average 4.4 (175 days) average 9.0 (253 days) 11.7 (>365 days) AC2-lf for 9 days 15.9 for 9 days for 68 days for 142 days average 8.3 (212 days) average 10.7 (304 days) 13.2 (>365 days) WPC1pre 10.8 5.0 11.5 18.4 27.7 2.7 1.5 3.5 3.9 10.4 WPC1post 16.7 7.9 17.5 13.5 20.1 4.9 2.7 5.9 5.9 9.3 WPC2pre 6.8 3.1 7.2 11.1 27.6 1.7 0.9 2.1 2.9 9.7 WPC2post 10.5 5.0 10.9 9.2 18.2 3.3 1.8 3.9 3.9 8 Very low risk: < 20 ng ΣPFAS10/L Very low risk: < 1 ng PFOA-eq/L Low risk: 20–50 ng ΣPFAS10/L Low risk: 1–2.5 ng PFOA-eq/L Medium risk: 50–100 ng ΣPFAS10/L Medium risk: 2.5–4 ng PFOA-eq/L High risk: 100–200 ng ΣPFAS10/L High risk: 4–5.5 ng PFOA-eq/L Very high risk: >200 ng ΣPFAS10/L Very high risk: > 5.5 ng PFOA-eq/L For the BF modelling, water flow boundary conditions (including daily river fluctuations as well as pumping rates) for the pre-climate change situations were taken from the year 2013, and for D2.4 – Guidance document on fate, transport and exposure for PMT’s 101 4. Catchment – river – riverbank filtration – drinking water the post-climate change situation from the year 2018 (see Annex G for more information). Table 14 shows that generally, the post-climate change (post-cc) scenarios predict higher ΣPFAS 10 and ΣPFAS 10 PFOA-equivalents concentrations as compared to the pre-climate change (pre-cc) scenarios. This means that according to this modelling study, climate change will generally increase health risks with regards to PFAS. This happens mostly because of the predicted increased length and severity of low–flow periods, which in turn increase river concentrations due to reduced dilution. However, when compared to the reference period (Ref), the post-cc baseline (BL) scenario predicts a decrease in PFAS concentrations in almost all cases. This is because the effects of climate change are counteracted by the reduction in industrial activity, especially the imminent closing of the PFAS production at Gendorf, which will lead to a very reduced PFAS load, particularly because of a reduction in ADONA emissions. Moreover, if water pollution control measures are implemented (WPC1 and WPC2), a definitive reduction in concentrations compared to the reference period is predicted, even for the post-cc situation. However, it is important to note that even in this scenario, risk levels for many areas are medium or higher according to the proposed (E)QS guidelines (EC COM(2022) 540 final). In terms of current DWD (EU DWD), only scenario WPC2 predicts a very low risk level at all three areas, although this depends on the contribution of contaminated background water. In Vienna, the ΣPFAS 10 as well as ΣPFAS 10 PFOA-equivalents concentrations decrease from the river Danube to the bank filtered (BF) water in all scenarios. This is due to mixing with ambient groundwater and sorption of PFAS to the aquifer material, leading to very low risk levels in all scenarios according to the current DWD. This has two further causes: (1) no inland background contamination is present, and (2) the riverbed at the Vienna BF site has a high degree of clogging. The effects of the first cause are evident, the second cause leads to reduced infiltration of river water into the aquifer. According to the thresholds in the draft of the (E)QS, risk levels are predicted to be low or medium in Vienna, but very high for the accidental spill of old AFFFs (AC-2), which will be elaborated upon below. In Budapest, PFAS concentrations are higher in groundwater than in river water in many of the simulated scenarios. This is due to mixing with inland groundwater from behind the PWs, shown to be contaminated (notably with PFOS, PFOA and GenX). The mixing ratio between river and ambient groundwater depends on water levels in the river Danube, the PWs and the background groundwater, which vary from day to day. This is why for the exact same river concentrations, different results can be observed in Surány versus Tahi, and this is also why the pre-cc and post-cc scenarios lead to different mixing ratios and thus, in some cases, to surprising results. For example, at Tahi, post-cc scenarios lead to a decrease of the ΣPFAS 10 concentrations from river to groundwater, while pre-cc scenarios lead to an increase. In terms of ΣPFAS 10 PFOA-equivalents, the groundwater at Tahi has the same exact concentration as the river concentration in all post-cc scenarios, showing that in these scenarios, the groundwater is almost exclusively affected by the river water. This shows the importance of discovering, monitoring and possibly treating inland background contamination, as well as studying the hydrological conditions at RBF sites. The results of the accidental spill (AC) scenarios are shown in Figure 4.17 to Figure 4.19 in more detail. The spills are assumed to happen over a period of 7 days, leading to a relatively short spike in PFAS concentrations in the river Danube at Vienna and Budapest. Because the spill is assumed to enter the river Danube in Linz, Austria, the levels at Vienna are higher than at Budapest. Furthermore, the peak concentrations for the low flow (lf) scenarios are higher than D2.4 – Guidance document on fate, transport and exposure for PMT’s 102 4. Catchment – river – riverbank filtration – drinking water for the high flow (hf) scenarios, due to a decrease in discharge in the river Danube, leading to reduced dilution. So, if a spill happens during low flow conditions, 4 times higher concentrations can be expected than during high flow conditions. Lastly, a spill of new AFFFs (AC-1) leads to about 10-15 times lower peak Σ10 PFAS concentrations and about and 40-100 times lower Σ10 PFOA-equivalents than old AFFFs (AC-2) in river water. In groundwater, these effects are somewhat attenuated (depending on the local hydrological situation) but still noticeable. The groundwater concentrations ultimately depend on the specific background concentration, transport time and subsurface characteristics at each well. These results can support water management during a PFAS spill event, as they can be used to make a quick estimate of the severity of the situation based on circumstantial information. Figure 4.17.: Development of a PFAS-peak from an accidental spill with new AFFF (top) and old AFFF (bottom) into river Danube close to Linz observed at the bank filtration site at Vienna. River concentrations in solid lines, groundwater concentrations in dotted lines. At the site in Vienna, it takes multiple months for the PFAS spill to infiltrate into the groundwater, during which dispersion leads to a flattening of the concentration peak. The riverbank at this site is so clogged that there is a 3-meter gradient between the water level in the Danube and the water level in the first groundwater monitoring well (1 m into the riverbank). This means that barely any water is able to flow though, which, together with sorption processes, leads to such extreme flattening of the peak. This is also the reason why PFAS concentrations do not cross the threshold of higher risk levels as defined in the current DWD in any spill scenario here (as can also be seen in Table 4.8), showing that although a clogged riverbank is not necessarily effective in regular situations (as discussed in section 4.6.3), it does prove a highly potent barrier in case of a temporary increase in river concentrations. Still, D2.4 – Guidance document on fate, transport and exposure for PMT’s 103 4. Catchment – river – riverbank filtration – drinking water groundwater PFAS concentrations do rise and peak after around 300 days. For the proposed (E)QS, AC1 results in low and medium risks for pre-cc and post-cc situations, respectively, while AC2 results in very high-risk levels for both cases. It is clear that the groundwater will be of such low quality for at least a year, and the best estimate is, that it will take at least three years for concentrations to return to the pre-spill levels, possibly longer. At Tahi, the concentrations of ΣPFAS 10 and ΣPFAS 10 PFOA-equivalents start increasing in the PW after about 25 days for the pre-cc scenarios and after about 50 days for the post-cc scenarios. This difference occurs because of the lower river level in the post-cc scenarios, leading to a lower gradient and thus slower groundwater flow. In the pre-cc scenarios, the increase of PFAS concentrations between 200 and 250 days after the spill does not happen due to infiltration of river water, but due to a sudden increase in the gradient between the pumping well and the monitoring well No. 3 due to falling water levels in both the river Danube and the pumping well, leading to an influx of PFAS from the inland source of contamination. This happens in all scenarios (not just the AC scenarios), highlighting the importance of a thorough understanding of the hydrogeological situation at the site of interest. Figure 4.18.: Development of a PFAS-peak from an accidental spill with new AFFF (top) and old AFFF (bottom) into river Danube close to Linz observed at the bank filtration site at Tahi, Budapest. River concentrations in solid lines, groundwater concentrations in dotted lines. Even though the effect is not as pronounced as in Vienna, mixing and sorption processes lead to a reduction and dispersion of the PFAS peaks at Tahi as well. This has the positive effect of a reduced maximum concentration compared to the river. The negative effect is that the PFAS concentrations in the pumping well are elevated for a longer time period. In practical terms, this means that the water does not meet the quality criteria for drinking water during D2.4 – Guidance document on fate, transport and exposure for PMT’s 104 4. Catchment – river – riverbank filtration – drinking water this time. As can be expected, risk levels are elevated higher and for a longer time if the spill contains old AFFF (AC-2) and if the flow conditions are lower (post-cc) (Table 4.8). Another interesting finding is that for the AC-1 scenarios, the ΣPFAS 10 is higher than the ΣPFAS 10 PFOA-equivalents, but for the AC-2 scenarios, this is the other way around. This is due to the high concentration of PFOS in old AFFFs, which has a high relative potency factor (RPF) due to its toxicity. Figure 4.19.: Development of a PFAS-peak from an accidental spill with new AFFF (top) and old AFFF (bottom) into river Danube close to Linz observed at the bank filtration site at Surány, Budapest. River concentrations in solid lines, groundwater concentrations in dotted lines. Due to Surány’s close proximity to Tahi, the PFAS peak in the river Danube was assumed to be the same. However, there are two important differences between Tahi and Surány. First, the hydrological situation is different, as the Surány pumping well is located further away from the riverbank and has a different pumping schedule than the Tahi pumping well. Generally, it can be stated that due to this difference, the spill has a smaller effect on groundwater concentrations in Surány than in Tahi, simply because less river water reaches the Surány pumping well. Second, the background contamination is a lot more pronounced at Surány, with higher concentrations of PFOS, PFOA and GenX. In the pre-cc scenarios, the significant increase in PFAS concentrations between 170 and 300 days after the spill is due to an increase in pumping rate, leading to a higher gradient between the background water and thus an influx of contaminated water. The same happens in the post-cc scenarios between 10 and 150 days after the spill. These effects are so pronounced that they tend to obscure the effects of the spill on the groundwater. D2.4 – Guidance document on fate, transport and exposure for PMT’s 105 5. Discussion and conclusions instance be calculated with the “SimpleBox - Aquatic Persistence Dashboard” based on physicalchemical characteristics. The approach presented to derive generic risk limits for soils shows that, depending on regional variations in geo(hydro)logical conditions, the high mobility of some PFAS could lead to strict requirements for materials applied on soil. 5.2. Pollution plumes in the soil-groundwater continuum For the soil-groundwater continuum, a novel model train coupling 1D-Hydrus, MODFLOW and MT3DMS to simulate PFAS fate and transport at watershed scale is presented. The model train accounts for the main physical, chemical and biological processes controlling the fate and transport of PFAS at this scale. For sorption and degradation reactions, several algorithms can be used allowing to select the most appropriate according to PFAS molecular properties and the characteristics of the simulated domain. A modelling framework describing the main steps required to build a model train has been also elaborated to facilitate the work of future modellers. Examples of the model train capabilities have been provided for two case studies, a PluriMetric Pilot experiment and a AFFF-polluted aquifer. The results of these modelling applications highlight the key role of identifying correctly the main physical and chemical processes controlling fate and transport of PFAS in the studied domain to build robust conceptual models. To increase model robustness, a thorough model calibration approach must be conducted, preferentially using time series measurements of PFAS concentration in pore solution in different locations of the contaminated site. Moreover, based on the results provided by the two simulated case studies, the key role of the unsaturated zone in the transfer and long-term migration of PFAS has been confirmed. If possible, a high frequency monitoring of PFAS concentration in the pore solution of the unsaturated zone should be performed as a sharp migration front can occur in this zone. For a broad range of PFAS sorption reactions in the unsaturated zone are expected to occur onto the soil-water and air-water interfaces, implying that PFAS sorption evolves according to water content. Therefore, numerical formalism more complex than linear isotherm needs to be used in the model developed for the soil-groundwater continuum, nonlinear and nonideal formalism should be used. Considering the key role of capillary fringe displacement on PFAS transport in the unsaturated zone, the model train seems very efficient to perform PFAS simulations as it can describe explicitly water flow and solute transport at the interface between unsaturated and saturated zones avoiding the main pitfall encountered in other numerical architecture. 5.3. Modelling of the catchment – river – riverbank filtration – drinking water continuum In the context of model applications for exposure assessment in the catchment – river – riverbank filtration – drinking water continuum, the guidance document demonstrates that the combination of stand-alone models in model trains expands the scope that can be covered. Model trains can combine individual models either in a complementary way or in a sequence. A complementary combination may either compare models of different complexity to find out which level of complexity (and associated effort) is needed to answer which questions or compare different D2.4 – Guidance document on fate, transport and exposure for PMT’s 112 5. Discussion and conclusions models with their different strengths and weaknesses in parallel to assess uncertainties, and/or use models for scenario evaluation according to their specific capabilities. The application of a model train incorporating the emission models MoRE and PPM demonstrated the capabilities of such a complementary approach based on the case of the Upper Danube catchment. Both models proved capable of predicting the instream concentration of 10 different PFAS based on emission estimates and produced with similar results in terms of exposure levels, pathway contributions and risk assessment of exceedance of drinking water or environmental quality standards in the surface water bodies of the catchment. Based on their specific strengths, both models contributed to the scenario assessment including climate scenarios, with MoRE being better able to represent scenarios with water protection measures and the PPM showing its main advantages in providing high temporal resolution results for accidental spill scenarios. A complementary application of a model train for comparing models with different levels of complexity for fate and transport modelling of bank filtration is presented for the Danube case study and related bank filtration sites. A 3D numerical reactive transport model proves to be the best choice for modelling the fate and transport of PFAS in a bank filtration setting. This is also true in situations where a steady-state approach is warranted, as the 3D cut-out models run sufficiently fast, so that computation time is not an issue. At the same time, the 3D numerical approach overcomes several conceptual shortcomings of the 1D approach, i.e. it can include (transverse) mixing and dispersion and (by extension) can better handle nonlinear (sorption) processes. For situations in which the complex dynamic interaction between surface water and groundwater is important, a transient 3D approach is indispensable. The results of the model train application combining the MoRE and the PPM emission models also provided the input to bank filtration modelling in a sequential combination of the model train for the upper Danube catchment. Such a sequential combination facilitates a broader application in terms of content and a combination of approaches at different spatial resolutions. Scenarios including catchment related aspects as water pollution control measures, climate scenarios and their impacts on exposure of surface waters have been combined with scenario assessment at bank filtrations sites and the risk of exceedance of limit values for groundwater or drinking water. The Berlin case demonstrates a sequential model train by combining an emission model of the city with a fate and transport model of the city´s surface waters. Main challenges for future improvements of exposure assessment for PFAS on catchment scale are: • the high relevance of legacy pollution from the use of fire-fighting foams or from old municipal landfills. Regarding the former, there is a need for more and better understanding of the extent of local groundwater pollution due to these applications (strongly linked to studies on the soil-groundwater continuum) and improved identification of contaminated sites. For landfills, the main challenges are related to the lack of a harmonised inventory of old municipal landfills, including information on size, age and type at international level. • the lack of robust, openly available information on the production, import-export and therefore use volumes of PFAS at national and EU level. A major effort is urgently needed D2.4 – Guidance document on fate, transport and exposure for PMT’s 113 5. Discussion and conclusions to provide this information, as it is decisive for sound environmental exposure assessment, not only for surface water and groundwater. • the huge number of PFAS, including thousands of substances beyond the ones discussed in this document. A reproducible and standardised analytical parameter for “total PFAS” or even for “total toxicity of PFAS” could be a very helpful in this context to address relevant PFAS in a combined way, if limits of quantifications are sufficiently low. • the limited scientific knowledge about the fate of PFAS in the environment. This includes knowledge of their partitioning between different phases (air, water, solids) and the transformations of so called “precursors” into stable “end products” like PFOA, PFOS and short chain substances. D2.4 – Guidance document on fate, transport and exposure for PMT’s 114 Bibliography AGES (2024). Analysis of soil parameters on behalf of the TU Vienna - Analysis orders 23142348001 and 23142348-002. Österreichische Agentur für Gesundheit und Ernährungssicherheit GmbH (AGES) (cit. on pp. 91, 92). Anderson, R. H., G. C. Long, R. C. Porter, and J. K. Anderson (2016). “Occurrence of select perfluoroalkyl substances at U.S. Air Force aqueous film-forming foam release sites other than fire-training areas: Field-validation of critical fate and transport properties”. In: Chemosphere 150, pp. 678–685. issn: 0045-6535. doi: 10.1016/j.chemosphere.2016.01.014 (cit. on p. 47). Bakker, M., V. Post, C. D. Langevin, J. D. Hughes, J. T. White, J. J. Starn, and M. N. Fienen (Sept. 2016). “Scripting MODFLOW Model Development Using Python and FloPy”. en. In: Groundwater 54.5, pp. 733–739. issn: 0017-467X, 1745-6584. doi: 10.1111/gwat.12413 (cit. on p. 58). Bakker, M., V. Post, C. D. Langevin, J. D. Hughes, J. White, J. Starn, and M. N. Fienen (2024). FloPy: Python package for creating, running, and post-processing MODFLOW-based models. Github code repository. doi: 10.5066/F7BK19FH. url: https://github.com/modflowpy/flopy (visited on 12/13/2024) (cit. on p. 58). Bear, J. (1979). Groundwater hydraulics. New York: McGraw-Hill (cit. on p. 75). Beegum, S., J. Šimůnek, A. Szymkiewicz, K. Sudheer, and I. M. Nambi (2018). “Updating the Coupling Algorithm between HYDRUS and MODFLOW in the HYDRUS Package for MODFLOW”. en. In: Vadose Zone Journal 17.1, p. 180034. issn: 1539-1663. doi: 10.2136/ vzj2018.02.0034 (cit. on p. 44). Behnisch, P., A. Sosnowska, E. Mombelli, and J. Kuckelkorn (2024). PROMISCES Deliverable D1.5 – Set of novel QSAR models/grouping/read-across and in vitro bioassay approaches predicting relevant toxicological endpoints for PFAS/iPM(T) chemicals. Tech. rep. version 2. doi: 10.3030/101036449. url: https://promisces.eu/Results/_/D1.5.pdf (cit. on p. 6). BWB (2024). “Unpublished measurement data from wastewater treatment plants, Berlin” (cit. on p. 79). Chorus, I. and M. Zessner (2021). “Assessing and controlling the risk of cyanobacterial blooms: Nutrient loads from the catchment”. eng. In: Toxic Cyanobacteria in Water. 2nd ed. United Kingdom: CRC Press, pp. 433–503. isbn: 978-0-367-53332-8. doi: 10.1201/9781003081449-7 (cit. on p. 68). Collenteur, R., G. Brunetti, and M. Vremec (2019). Phydrus: Python implementation of the HYDRUS-1D unsaturated zone model (cit. on p. 58). EC COM(2019)640 final (Dec. 2019). Communication from the Commission to the European Parliament, the European Council, the Council, the European Economic and Social Committee and the Committee of the Regions: The European Green Deal. en. Publisher: Publications Office of the European Union. url: https://op.europa.eu/en/publication-detail/-/publication/ b828d165-1c22-11ea-8c1f-01aa75ed71a1/language-en (visited on 01/15/2025) (cit. on p. 21). D2.4 – Guidance document on fate, transport and exposure for PMT’s 115 Bibliography DELTARES (2025). D-Emissions, Water quality and aquatic ecology modelling suite, Assessment tool for the emitted pollution at catchment scale, user manual. Version 7.00. url: https: //content.oss.deltares.nl/delft3d/D-Emissions_User_Manual.pdf (cit. on p. 70). Derx, J., J. Komma, P. Reiner, J. Vierheilig, D. Savio, R. Sommer, A. K. T. Kirschner, C. Frick, R. Linke, A. H. Farnleitner, and A. P. Blaschke (2021). “Using hydrodynamic and hydraulic modelling to study microbiological water quality issues at a backwater area of the Danube to support decision-making”. en. In: Österreichische Wasserund Abfallwirtschaft 73.11-12, pp. 482–489. issn: 0945-358X, 1613-7566. doi: 10.1007/s00506-021-00797-7 (cit. on p. 92). EU DWD. Directive (EU) 2020/2184 of the European Parliament and of the Council of 16 December 2020 on the quality of water intended for human consumption (recast).url: https://eur-lex.europa.eu/eli/dir/2020/2184/oj/eng. (Cit. on pp. 15, 87, 90, 100–102). European Comission (EC) (2012). Common Implementation Strategy for the Water Framework Directive (2000/60/EC) Guidance Document No. 28 Technical Guidance on the Preparation of an Inventory of Emissions, Discharges and Losses of Priority and Priority Hazardous Substances. eng. Tech. rep. European Comission (EC). doi: 10.2779/2764. url: https: //data.europa.eu/doi/10.2779/2764 (visited on 01/15/2025) (cit. on pp. 67, 68). EC C(2022) 9383 final (2022). ANNEXES to the Commission Delegated Regulation amending Regulation (EC) No 1272/2008 as regards hazard classes and criteria for the classification, labelling and packaging of substances and mixtures. Reference information missing. url: https: //environment.ec.europa.eu/system/files/2022-12/Annexes%20to%20the%20Delegated% 20Regulation.pdf (cit. on p. 28). EC COM(2020) 98 final (July 2020). Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions – A new Circular Economy Action Plan for a cleaner and more competitive Europe. en. Publisher: Publications Office of the European Union. url: https://op.europa.eu/ en/publication-detail/-/publication/d33bd445-1930-11eb-b57e-01aa75ed71a1/language-en (visited on 01/15/2025) (cit. on p. 21). EC COM(2022) 540 final (2022). Proposal For A Directive Of The European Parliament And Of The Council amending Directive 2000/60/EC establishing a framework for Community action in the field of water policy, Directive 2006/118/EC on the protection of groundwater against pollution and deterioration and Directive 2008/105/EC on environmental quality standards in the field of water policy.url: https://environment.ec.europa.eu/publications/proposalamending-water-directives_en (visited on 10/26/2022) (cit. on pp. 16, 17, 87, 90, 100–102). EC COM(2021) 400 final (Oct. 2021). Opinion of the European Economic and Social Committee on the communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions – Pathway to a Healthy Planet for All – EU Action Plan: ‘Towards Zero Pollution for Air, Water and Soil’. en. (OJ C, C/105, 04.03.2022, p. 143). url: https://eur-lex.europa.eu/legalcontent/EN/TXT/?uri=CELEX:52021AE2629 (visited on 01/15/2025) (cit. on p. 21). ECHA (2024). REACH Registration Dossier: 6:2 fluorotelomer sulfonamide alkylbetanin.url: https://chem.echa.europa.eu/100.047.300 (cit. on p. 29). EFSA, C., D. Schrenk, M. Bignami, L. Bodin, J. K. Chipman, J. del Mazo, B. Grasl-Kraupp, C. Hogstrand, et al. (2020). “Risk to human health related to the presence of perfluoroalkyl substances in food”. en. In: EFSA Journal 18.9. EFSA Panel on Contaminants in the Food Chain (EFSA CONTAM Panel). issn: 1831-4732. doi: 10.2903/j.efsa.2020.6223 (cit. on p. 37). D2.4 – Guidance document on fate, transport and exposure for PMT’s 116 Bibliography 2000/60/EC. DIRECTIVE 2000/60/EC OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 23 October 2000 establishing a framework for Community action in the field of water policy” or, in short, the EU Water Framework Directive.url: http: //data.europa.eu/eli/dir/2000/60/oj. (Cit. on p. 66). European Chemical Agency (ECHA) (2016). Guidance on information requirements and chemical safety assessment Chapter R.16: Environmental Exposure Estimation. Version 3.0 February 2016. en. ECHA. doi: 10.2823/515110. url: https://echa.europa.eu/documents/10162/ 17224/information_requirements_r16_en.pdf/ (cit. on p. 28). Fuchs, S., M. Kaiser, L. Kiemle, S. Kittlaus, S. Rothvoß, S. Toshovski, A. Wagner, R. Wander, T. Weber, and S. Ziegler (Mar. 2017). “Modeling of Regionalized Emissions (MoRE) into Water Bodies: An Open-Source River Basin Management System”. en. In: Water 9.4, p. 239. issn: 2073-4441. doi: 10.3390/w9040239. url: https://www.mdpi.com/2073-4441/9/4/239 (visited on 12/13/2024) (cit. on p. 70). Goedecke, M., L. Haag, and J. H.Gerstenberg (2019). Wasserhaushaltsmodell Berlin ABIMO 3.2 - Handreichung für Anwendende Technische Anleitung zur Aufbereitung von Datengrundlagen sowie Dokumentation von Methoden und Berechnungsergebnissen des auf Berliner Verhältnisse angepassten blockbezogenen Niederschlags - Abflussmodell ABIMO der Bundesanstalt für Gewässerkunde, 2006 bis 2019 (cit. on p. 79). Groot, H., A. Sosnowska, W. Peijnenburg, J. Meesters, A. Wintersen, V. Zhiteneva, N. Devau, J. Valstar, et al. (2025). “PROMISCES D2.3 Toolbax fate & transport modelling of PMTs in the environment”. in preparation. doi: 10.3030/101036449. url: https://cordis.europa.eu/ project/id/101036449/results (cit. on pp. 21, 94–96). Gupta, H. V., H. Kling, K. K. Yilmaz, and G. F. Martinez (2009). “Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling”. In: Journal of Hydrology 377.1-2, pp. 80–91. issn: 00221694. doi: 10.1016/j. jhydrol.2009.08.003 (cit. on p. 16). Handl, S., E. Mayr, and R. Perfler (June 2017). “Brunnenmonitoring – Beurteilung von Einflussfaktoren auf die Leistungsfähigkeit bei Uferfiltratbrunnen”. de. In: Österreichische Wasserund Abfallwirtschaft 69.5, pp. 240–246. issn: 1613-7566. doi: 10.1007/s00506-0170385-0. url: https://doi.org/10.1007/s00506-017-0385-0 (visited on 01/15/2025) (cit. on p. 65). Harbaugh, A. W. (2005). “MODFLOW-2005 : the U.S. Geological Survey modular ground-water model–the ground-water flow process”. en. In: Techniques and Methods. Number: 6-A16. issn: 2328-7055. doi: 10.3133/tm6A16. url: https://pubs.usgs.gov/publication/tm6A16 (visited on 01/15/2025) (cit. on p. 78). Held, T. and M. Reinhard (2020). Remediation management for local and wide-spread PFAS contaminations. en. Umweltbundesamt. url: https://www.umweltbundesamt.de/en/ publikationen/remediation-management-for-local-wide-spread-pfas (visited on 01/15/2025) (cit. on p. 58). Høisæter, Å., A. Pfaff, and G. D. Breedveld (2019). “Leaching and transport of PFAS from aqueous film-forming foam (AFFF) in the unsaturated soil at a firefighting training facility under cold climatic conditions”. In: Journal of Contaminant Hydrology 222, pp. 112–122. issn: 0169-7722. doi: 10.1016/j.jconhyd.2019.02.010 (cit. on p. 58). Hughes, J. D., C. D. Langevin, S. R. Paulinski, J. D. Larsen, and D. Brakenhoff (2023). “FloPy Workflows for Creating Structured and Unstructured MODFLOW Models”. en. In: Groundwater 62.1, pp. 124–139. issn: 1745-6584. doi: 10.1111/gwat.13327 (cit. on p. 58). D2.4 – Guidance document on fate, transport and exposure for PMT’s 117 Bibliography ITRC (2015). Integrated DNAPL Site Characterization and Tools Selection (ISC-1).url: https://projects.itrcweb.org/DNAPL-ISC_tools-selection/ (cit. on p. 77). Li, L., A. Sangion, F. Wania, J. M. Armitage, L. Toose, L. Hughes, and J. A. Arnot (Dec. 2021). “Development and Evaluation of a Holistic and Mechanistic Modeling Framework for Chemical Emissions, Fate, Exposure, and Risk”. In: Environmental Health Perspectives 129.12. Publisher: Environmental Health Perspectives, p. 127006. doi: 10.1289/EHP9372. url: https://ehp.niehs.nih.gov/doi/10.1289/EHP9372 (visited on 01/15/2025) (cit. on p. 24). Liu, M., E. Saracevic, S. Kittlaus, T. Oudega, A. Obeid, Z. Nagy-Kovács, B. László, N. Krlovic, Z. Saracevic, G. Lindner, G. Rab, J. Derx, O. Zoboli, and M. Zessner (2023). “PFAS-Belastungen im Einzugsgebiet der oberen Donau”. de. In: Österreichische Wasserund Abfallwirtschaft 75.9, pp. 503–514. issn: 1613-7566. doi: 10.1007/s00506-023-00973-x (cit. on p. 79). Lyu, X., F. Xiao, C. Shen, J. Chen, C. M. Park, Y. Sun, M. Flury, and D. Wang (2022). “Perand Polyfluoroalkyl Substances (PFAS) in Subsurface Environments: Occurrence, Fate, Transport, and Research Prospect”. en. In: Reviews of Geophysics 60.3, e2021RG000765. issn: 1944-9208. doi: 10.1029/2021RG000765 (cit. on p. 47). Meesters, J.A.J. (2024). PROMISCES D2.6 : An improved SimpleBox model for improved environmental risk assessment and life cycle impact assessment.doi: 10.3030/101036449 (cit. on pp. 28, 29, 31, 32). Nagy-Kovács, Z., J. Davidesz, K. Czihat-Mártonné, G. Till, E. Fleit, and T. Grischek (Feb. 2019). “Water Quality Changes during Riverbank Filtration in Budapest, Hungary”. en. In: Water 11.2, p. 302. issn: 2073-4441. doi: 10.3390/w11020302 (cit. on p. 92). Neumann, M. and I. Schliebner (Nov. 2019). Protecting the sources of our drinking water: The criteria for identifying persistent, mobile and toxic (PMT) substances and very persistent and very mobile (vPvM) substances under EU Regulation REACH (EC) No 1907/2006. en. Umweltbundesamt. url: https://www.umweltbundesamt.de/en/publikationen/protectingthe-sources-of-our-drinking-water-the (cit. on pp. 76, 77). Nguyen, T. M. H., J. Bräunig, K. Thompson, J. Thompson, S. Kabiri, D. A. Navarro, R. S. Kookana, C. Grimison, C. M. Barnes, C. P. Higgins, M. J. McLaughlin, and J. F. Mueller (Dec. 2020). “Influences of Chemical Properties, Soil Properties, and Solution pH on Soil–Water Partitioning Coefficients of Perand Polyfluoroalkyl Substances (PFASs)”. en. In: Environmental Science & Technology 54.24, pp. 15883–15892. issn: 0013-936X, 1520-5851. doi: 10.1021/acs.est.0c05705 (cit. on pp. 91, 92). Obeid, A. A., T. J. Oudega, O. Zoboli, C. Gundacker, A. P. Blaschke, M. Zessner, E. Saracevic, N. Devau, et al. (2023). “The occurrence and persistence of PFAS at riverbank filtration sites in the Upper Danube basin”. de. In: Österreichische Wasserund Abfallwirtschaft 75.9-10, pp. 515–527. issn: 0945-358X, 1613-7566. doi: 10.1007/s00506-023-00974-w (cit. on pp. 91, 92, 95). Obeid, A. A., T. J. Oudega, O. Zoboli, C. Gundacker, A. P. Blaschke, M. Zessner, E. Saracevic, N. Devau, et al. (2025). “The effect of different molecular physicochemical properties on PFAS transport and sorption behaviour in saturated porous media”. In preparation (cit. on p. 92). Oudega, T. J., A. A. Obeid, N. Devau, O. Zoboli, C. Gundacker, A. P. Blaschke, M. Zessner, E. Saracevic, et al. (2024). “PFAS transport and retention during riverbank filtration and in saturated columns”. In: River Basins Conference, Budapest.doi: 10.3311/rb2024 (cit. on p. 73). D2.4 – Guidance document on fate, transport and exposure for PMT’s 118 Bibliography EUSES 2024 (2024). Presentation on theme: "USES guia del USES 4.0. EUSES (European Union System for the Evaluation of Substances) Multimedia.url: https://slideplayer.com/ slide/4857718/ (visited on 01/15/2025) (cit. on p. 26). SimpleBox Website (2024). http://www.rivm.nl/SimpleBox (cit. on p. 27). Schepens, M., J. te Biesebeek, J. Hartmann, N. van der Aa, R. Zijlstra, and P. Boon (2023). Risk assessment of exposure to PFAS through food and drinking water in the Netherlands. Tech. rep. Rijksinstituut voor Volksgezondheid en Milieu RIVM. doi: 10.21945/RIVM-2023-0011. url: https://rivm.openrepository.com/handle/10029/626814 (visited on 01/15/2025) (cit. on p. 37). Schumacher, F. (2023). BIBER (Berechnungsund Informationssystem Berliner Oberflächengewässer) Stationäre Berechnung der Wasserstandsund Durchflussverhältnisse im Berliner Gewässersystem bei niedrigen (MNQ) und mittleren (MQ) Durchflüssen. Tech. rep. Ingenieurbüro für Wasser und Umwelt (cit. on p. 80). Simunek Jirka, J., M. Šejna, and M. Van Genuchten (Jan. 2013). The HYDRUS-1D Software Package for Simulating the One-Dimensional Movement of Water, Heat, and Multiple Solutes in Variably-Saturated Media. (Cit. on pp. 78, 94). Smits, J. G. C. and J. K. L. van Beek (July 2013). “ECO: A Generic Eutrophication Model Including Comprehensive Sediment-Water Interaction”. en. In: PLOS ONE 8.7. Publisher: Public Library of Science, e68104. issn: 1932-6203. doi: 10.1371/journal.pone.0068104. url: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0068104 (visited on 01/15/2025) (cit. on p. 70). Sosnowska, A., T. Puzyn, W. Peijnenburg, and P. Wassenaar (2024). PROMISCES Deliverable D2.1 - Toolbox improved in silico models for identification of PMT properties. Tech. rep. QSARLab and RIVM. doi: 10.5281/zenodo.14800915. url: https://promisces.eu/Results/_/ D2.1.pdf (cit. on pp. 25, 29, 30). Swartjes, F. A., ed. (2011). Dealing with Contaminated Sites: From Theory towards Practical Application. en. Dordrecht: Springer Netherlands. isbn: 978-90-481-9757-6. doi: 10.1007/97890-481-9757-6. url: https://link.springer.com/10.1007/978-90-481-9757-6 (visited on 01/15/2025) (cit. on p. 34). Szabó, B., M. Weynants, and T. K. D. Weber (Jan. 2021). “Updated European hydraulic pedotransfer functions with communicated uncertainties in the predicted variables (euptfv2)”. English. In: Geoscientific Model Development 14.1. Publisher: Copernicus GmbH, pp. 151–175. issn: 1991-959X. doi: 10.5194/gmd-14-151-2021. url: https://gmd.copernicus.org/articles/ 14/151/2021/ (visited on 01/15/2025) (cit. on p. 36). Togola, A., E. Saracevic, T. Hensel, B. Idjaton, and F. Zietzschmann (2024). PROMISCES Deliverable D1.3 – Methods for global organic fluorinated content (TOP, TOF/AOF/EOF) for relevant matrices. Tech. rep. doi: 10.3030/101036449. url: https://cordis.europa.eu/ project/id/101036449/results (cit. on p. 6). US EPA, O. (2024). CompTox Chemicals Dashboard. en. Data and Tools. url: https://www. epa.gov/comptox-tools/comptox-chemicals-dashboard (visited on 01/15/2025) (cit. on pp. 29, 30). Verbruggen, E., P. Wassenaar, and C. Smit (2017). Water quality standards for PFOA : A proposal in accordance with the methodology of the Water Framework Directive 2017-0044. RIVM. doi: 10.21945/RIVM-2017-0044. url: http://hdl.handle.net/10029/620918 (cit. on pp. 37, 39). D2.4 – Guidance document on fate, transport and exposure for PMT’s 119 Bibliography Vermeulen, P. T. M., G. M. C. M. Janssen, and T. Kroon (2024). “Efficient Model Calibration Using Submodels”. en. In: Water Resources Research 60.11. issn: 1944-7973. doi: 10.1029/ 2023WR036441 (cit. on p. 78). Verseveld, W. J. van, A. H. Weerts, M. Visser, J. Buitink, R. O. Imhoff, H. Boisgontier, L. Bouaziz, D. Eilander, M. Hegnauer, C. ten Velden, and B. Russell (Apr. 2024). “Wflow_sbm v0.7.3, a spatially distributed hydrological model: from global data to local applications”. English. In: Geoscientific Model Development 17.8, pp. 3199–3234. issn: 1991-959X. doi: 10.5194/gmd-17-3199-2024 (cit. on p. 70). Wang, Z., G. W. Walker, D. C. G. Muir, and K. Nagatani-Yoshida (Mar. 2020). “Toward a Global Understanding of Chemical Pollution: A First Comprehensive Analysis of National and Regional Chemical Inventories”. en. In: Environmental Science & Technology 54.5, pp. 2575–2584. issn: 0013-936X, 1520-5851. doi: 10.1021/acs.est.9b06379 (cit. on p. 24). West, M. R., B. H. Kueper, and M. J. Ungs (Mar. 2007). “On the Use and Error of Approximation in the Domenico (1987) Solution”. en. In: Groundwater 45.2, pp. 126–135. issn: 0017-467X, 1745-6584. doi: 10.1111/j.1745-6584.2006.00280.x (cit. on p. 75). Wicke, D., A. Matzinger, and P. Rouault (2015). Abschlussbericht Relevanz organischer Spurenstoffe im Regenwasserabfluss Berlins. Tech. rep. Kompetenzzentrum Wasser Berlin (cit. on p. 79). Wintersen, A., L. Oste, R. van der Meiracker, P. van Breemen, G. Roskam, and J. Spijker (2020). Verschil in uitloging van PFAS uit grond en bagger. Tech. rep. Rijksinstituut voor Volksgezondheid en Milieu (RIVM). doi: 10.21945/RIVM-2020-0102. url: https://rivm. openrepository.com/handle/10029/623865 (visited on 01/15/2025) (cit. on p. 39). Zheng, C. and P. Wang (Dec. 1999). MT3DMS: A modular three-dimensional multispecies transport model for simulation of advection, dispersion, and chemical reactions of contaminants in groundwater systems; documentation and user’s guide. Tech. rep. Contract report SERDP99-1. Vicksburg, Mississippi, USA: United States Army Engineer Research and Development Center, p. 220 (cit. on p. 78). D2.4 – Guidance document on fate, transport and exposure for PMT’s 120 A. Substance properties to be used in the SB-AP Dashboard Authored by: Johannes AJ Meesters Dutch National Institute for Public Health and Environment (RIVM) Table A.1.: Derived biodegradability rate constants in water based on REACH Registration Dossier Information Substance short name and CAS number Degradability Test Description(s) of test outcome(s) Interpreted minimum degradation rate constant (log( s-1) ) Interpreted maximum degradation rate constant (log( s-1) ) Registration dossier cC604 CAS: 119093127-1 OECD 301 A No significant degradation after 28-day DOC Die-Away Test Log (1/100 years) = - 9.5 log( s-1) Log (-ln(10.001)/28 days) = - 9.38 log( s-1) ECHA 2022b 6:2 FTSA CAS: 27619-97-2 OECD 301 B After 28 days of incubation, the substance degradation reached a maximum mean of 42% Log (-ln(10.42)/28 days) = - 6.65 log( s-1) Log (-ln(10.42)/28 days) = - 6.7 log( s-1) ECHA 2019 PFBS CAS: 375-73-5 Read across from structural analogue Potassium1,1,2,2,3,3, 4,4,4nonafluorobutane-1sulphonate, CAS 29420-49-3 Given 14% degradation after 28 days, the structural analogue is not readily biodegradable according to OECD criteria Log (-ln(10.14)/28 days) = - 7.2 log( s-1) Log (-ln(10.14)/28 days) = - 7.2 log( s-1) ECHA 2018b D2.4 – Guidance document on fate, transport and exposure for PMT’s 121 B. PluriMetric Pilot experiments was kept stable at 1m depth below the surface of the PMP. Therefore, the thicknesses of the unsaturated and saturated zones were of 1 m and 2 m, respectively. Furthermore, controlled water flow and mass transport boundary conditions are maintained during the duration of the experiment. Concerning water flow, three boundary conditions have been fixed. Two boundary conditions in the saturated zone: a hydraulic barrier on one side of the PMP maintaining a constant hydraulic head and a pumping well located on the opposite side. The pumping rate is maintained at a constant flow rate of 4x10 -2 m 3 /h during all the duration of the experiment to maintain steady-state conditions in the saturated zone. The two other sides of the PMP have no flow boundary conditions. Pressure transducers are connected to the pumps providing water in the hydraulic barrier to maintain constant hydraulic head at 1 m depth below the surface. Water recharge and PFAS injection was performed at the top of the PMP using a sprinkler system deployed on a square of 2.25 m 2 allowing imbibition of the unsaturated zone. The surface of the infiltration square corresponds to nearly 10% of the total surface at the top of the PMP. The infiltration square is located closer to the side of the PMP where hydraulic head (1.5 m between the center of the square and this side of the PMP) is kept constant than the pumping well (the center of the square is 3 m downstream of the pumping well). The AFFF used during the PMP experiment mainly contains PFAS with 8 carbon chain length. The main PFAS found in the AFFF are 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam (1525.0 mg/L, 3409.0 mg/L, 780 mg/L, respectively) and, in a lesser extent, PFHxA, 8:2 FTSA, 4:2 FTSA and 10:2 FTSA (8.9 mg/L, 3.8 mg/L, 3.4 mg/L, 1.7 mg/L). Other chemical compounds as also part of the AFFF formulation such as polyethylene glycol or ionic salts (SO 42and NH 4+ ). This AFFF formulation is similar to the one reported in literature for products developed since 2000. Homogeneity and stability of the product has been assessed by performing multiple sampling and analysis on the stored mother solution. The monitoring set up in PMP is based on probes and sampling device allowing us to the monitor water table depth, soil moisture and soil water potential in the vadose zones as well as sample water in vadose and saturated zones. The water table level has been monitored by using automatic pressure transducer installed in 9 piezometers. Capillary pressure and water content in the vadose zone has been monitored using 15 tensiometers and 15 TDR. To collect water samples, 80 suctions cups were positioned along the main flow lines in different locations in the PMP and at different depths from -0.5m to -2.5m. Water samples were also collected in the 9 piezometers. Sensors and probes are distributed according to a regular grid at the surface and in the depth of the PMP. This grid has been selected to ensure robust calculations of PFAS distribution in the PMP at the end of the experiment. The following physical/chemical properties have been measured on all the water samples: pH, redox, temperature, electrical conductivity. On a more limited number of samples, additional analyses have been conducted, bromide concentrations and total organic fluoride concentrations. On a subset of these latter samples, targeted PFAS analyses have been done to quantify the concentration of each PFAS present in the AFFF. More details concerning the protocol used to perform these chemical analyses can be found in WP1 deliverables. The experiment was designed to reproduce a real AFFF contamination of a SGW continuum. More precisely, PFAS migration from the infiltration square at the top of the PPM to the pumping well located 3m downstream via vertical solute transport through the unsaturated zone and horizontal solute transport in the saturated zone was targeted. This migration pattern has been indeed encountered in several real AFFF contaminated site. The experiment has been D2.4 – Guidance document on fate, transport and exposure for PMT’s 128 B. PluriMetric Pilot experiments conducted during 11 months, with specifically 8 months dedicated to monitor PFAS migration. During this time length, the experiment has been divided in the three following stage to achieve this goal: • Baseline stage. This stage was devoted to acquiring a baseline of the PMP prior to AFFF contamination. The flow rate in the saturated zone was kept at the fixed value aforementioned. Weekly recharge events were performed by infiltrating 120L of PFAS-free water on the infiltration square at the top of the PMP. The time length for this stage is 2.5 months (from mid-February 2024 to mid-April 2024). Prior to this stage, a stabilisation stage of 4 months was applied. A sampling campaign has been conducted at the beginning and the end of this stage. Water was sampled in all sampling points in the PMP. • PFAS infiltration stage. This stage was conducted during three weeks (from mid-April 2024 to mid-May 2024). A diluted solution of AFFF (3% w/w) was infiltrated from the injection square. The total volume of input solution is equal to 1.6 m 3 . This volume has been selected to reach PFAS concentration in pore solution and sorbed PFAS concentrations in the range of the concentrations measured in highly contaminated sites mentioned in literature. To prevent saturation of the unsaturated zone, the total volume of diluted AFFF solution was infiltrated in 14 infiltration periods. During each infiltration period, 120 L of diluted AFFF solution was infiltrated during 3h. A non-reactive tracer (Br) was also added to the input solution to characterize the transport properties of the PMP during PFAS migration. The flow rate in the saturated zone is maintained at the value fixed in the first stage. A monitoring campaign was conducted after each PFAS injection step at the sampling points located below the infiltration square and in its vicinity. • PFAS leaching stage. This stage was performed to induce PFAS migration from the infiltration square to the unsaturated and saturated zones as it can be induced by natural recharge on a real AFFF-contaminated. During this stage, PFAS-free water was infiltrated on a weekly basis during 5 months (from mid-May to October 2024) in the injection square. The infiltrated volume is equal to 120L. Each week, water samples were collected in all sampling points in the PMP. The prior step before building a model of the PMP experiment was to conduct a comprehensive interpretation of the results, notably in-depth data analyses. This prior step before building the model is a cornerstone to be able to build a correct conceptual model of the PMP experiment. The following conclusions were drawn based on the results obtained from the measured parameters during the experiment: • The concentrations in PFAS are lower in pore water in both unsaturated and saturated zones than in the diluted AFFF solution. In the pore solution, the decrease is more marked for 6:2 FTAB and 6:2 FTSAam than for 6:2 FTSA. This latter compound has become predominant in the dissolved PFAS (in the input solution, 6:2 FTAB is predominant). The magnitude of the decrease is, in mean during all the experiment, equal to 50%, 78%, 98% for 6:2 FTSA, 6:2 FTAB and 6:2 FTSAam, respectively. These decreases of PFAS concentrations in the pore water suggest that sorption reactions onto the AWI and SWI had a strong impact on PFAS migration in the unsaturated zone. In the saturated zone, in addition to the sorption reactions, dilution of the high PFAS pore solution coming D2.4 – Guidance document on fate, transport and exposure for PMT’s 129 B. PluriMetric Pilot experiments from the unsaturated zone below the infiltration square by the PFAS-free pore solution injected in the PMP to maintain groundwater flow can also explained the decrease of PFAS concentration in the saturated zone • The affinity of the 3 main PFAS to the AWI and/or SWI differs. Based on their affinity to these two interfaces, the PFAS compounds can be ranked as following 6:2 FTSAam > 6:2 FTAB > 6:2 FTSA, Hence, the residence time in the geomedia is lower for 6:2 FTSA than 6:2 FTAB and even more for 6:2 FTSAam. This interpretation is supported by the fact that 6:2 FTSA and 6:2 FTAB have been measured in pore solution during all the experiment. In the sampling points located downstream of the infiltration square in the saturated zone, only 6:2 FTSA had been found in pore water. Concentrations of 6:2 FTSAam were close to null in pore water in both unsaturated and saturated zones during all the experiment. • In all the sampling points, concentrations of 6:2 FTSA change with time according to a similar pattern. This pattern can be divided into three periods: concentration of 6:2 FTSA is close to null, then an increase of the concentration is measured followed by a decrease of the concentration. The duration of each one of these periods as well as the magnitude of the increase or decrease vary according to the locations in space and depth of the sampling points. These results highlight that inter-related processes including advection-dispersion and sorption reactions onto the SWI and/or AWI as well as mixing between water with different PFAS concentration impact the change in 6:2 FTSA concentrations during the experiment. • For the 6:2 FTSA breakthrough curves measured in the sampling points below the infiltration square (in both unsaturated and saturated zone), 6:2 FTSA concentration increased drastically during the infiltration stage. The higher concentrations were measured in the sampling points located just below the capillary fringe in the saturated zone, illustrating that 6:2 FTSA is mainly stored in this transition zone between the unsaturated and saturated zone. Then, the PFAS concentration decreases at the start of the leaching phase. The decrease was more marked in the sampling points located in the saturated zone than unsaturated zone, suggesting that a higher amount of 6:2 FTSA was sorbed in this latter zone. To explain this trend, a hypothesis that sorption reactions onto the AWI had played a key role in 6:2 FTSA fate and transport in the unsaturated zone during the experiment can be claimed. The intensity of the sorption seems higher in the upper part of the unsaturated zone than in the deeper part. • As the distance between the infiltration square and the sampling point increases, the time delay between the start of the diluted AFFF infiltration and the rising of 6:2 FTSA concentration in the pore water increases. Furthermore, the increase is lower in the sampling points located far from the infiltration square. These results suggest that reactive processes seem to play a lower role than transport processes as the distance between the infiltration square increases. • The trends of the breakthrough curves measured for 6:2 FTAB and 6:2 FTSAam were similar to the one aforementioned for 6:2 FTSA but the rising of the concentration was lower. Based on these results, one can hypothesis that the same physical and chemical processes have controlled the fate and transport of 6:2 FTSA, 6:2FTAB and 6:2 FTSAam D2.4 – Guidance document on fate, transport and exposure for PMT’s 130 B. PluriMetric Pilot experiments during the experiment only the magnitude of the processes, notably sorption reactions onto the SWI and/or AWI had differed. • No significant changes in the concentrations of the PFAS resulting from the degradation reaction of the precursors contained in the diluted AFFF solution infiltrated at the top of the PMP have been measured, suggesting that no degradation reactions occurred during the experiment. These results are in line with previous experimental results obtained either on laboratory conditions or on real AFFF-contaminated sites. • Interpretations of the breakthrough curves measured for the non-reactive tracer in the sampling points located below the infiltration square indicate that evaporation process occurs during the PFAS-infiltration and -leaching stages. This hypothesis needs to be formulated to explain the increase of Br concentration in the pore water above the value of the infiltrated solution. • The hydraulic properties in the unsaturated and saturated zones below the infiltration square change drastically during the PFAS-infiltration and -leaching stages as indicated by the measurements of water content and capillary pressure from few centimetres below the surface to 3 m depth. This change can be explained by foam formation during diluted AFFF infiltration stage. Although infiltration flow rate was maintained low to prevent foam formation, foam had still been produced in pore network during this stage. By expanding, the PFAS-induced foam had modified the connection between the pores, leading to create preferential flow lines in the soil column below the infiltration square. This change in pore network is in line with the quick increase of PFAS concentration in sampling point located both in the unsaturated and saturated zones below the infiltration square. PFAS-induced foam seems stable with time the capillary pressure and water content in response to the infiltration with the same water rate are identic during and after AFFF-infiltration stage. • AFFF infiltration in the geomedia seems also to change hydraulic properties in the unsaturated and saturated zones, notably permeability. In the unsaturated zone, changes in capillary pressure and water content in response to water infiltration were faster during and after AFFF infiltration stages compared to measurements perform during the baseline stage. In the saturated zone, interpretations of AFFF infiltration on hydraulic properties are more difficult to characterize but are also suspected. B.3. Hydraulic, solute transport and reactive properties for simulations of the CS#6 Table B.1.: Hydraulic properties for each layer of the 1D-vertical model and groundwater flow models Layers Qr Qs Alpha (m-1) n Ks (m/day) l 1D-Hydrus Layer 1 0.045 0.38 14.5 2.68 0.31 0.5 D2.4 – Guidance document on fate, transport and exposure for PMT’s 131 B. PluriMetric Pilot experiments Layers (thickness) Qr Qs Alpha (m-1) n Ks (m/day) l 1D-Hydrus Layer 2 0.045 0.38 14.5 2.68 0.41 0.5 1D-Hydrus Layer 3 0.045 0.38 14.5 2.68 0.41 0.5 1D-Hydrus Layer 4 0.045 0.25 14.5 2.68 0.41 0.5 1D-Hydrus Layer 5 0.045 0.2 14.5 2.68 0.41 0.5 1D-Hydrus Layer 6 0.045 0.2 14.5 2.68 0.41 0.5 MODFLOW/MT3DMS Layer 1 below infiltration square - 0.2 - - 0.31 - MODFLOW/MT3DMS Layers 2-10 below infiltration square - 0.2 - - 0.31 - MODFLOW/MT3DMS Layer 1 outside infiltration square - 0.38 - - 0.31 - MODFLOW/MT3DMS Layers 2-10 outside infiltration square - 0.38 - - 0.31 - Table B.2.: Solute transport and reactive properties for each layer of the 1D-vertical model and groundwater flow models Layers Bulk density (kg/m3) Disp. L (m) Disp. T. (m) 1D-Hydrus Layer 1 1700 0.2 - 1D-Hydrus Layer 2 1700 0.2 - 1D-Hydrus Layer 3 1700 0.2 - 1D-Hydrus Layer 4 1700 0.1 - 1D-Hydrus Layer 5 1700 0.05 - 1D-Hydrus Layer 6 1700 0.05 - MODFLOW/MT3DMS Layer 1 below infiltration square 1700 0.05 0.05 MODFLOW/MT3DMS Layers 2-10 below infiltration square 1700 0.05 0.05 D2.4 – Guidance document on fate, transport and exposure for PMT’s 132 B. PluriMetric Pilot experiments Layers (thickness) Bulk density (kg/m3) Disp. L (m) Disp. T. (m) MODFLOW/MT3DMS Layer 1 outside infiltration square 1700 0.2 0.2 MODFLOW/MT3DMS Layers 2-10 outside infiltration square 1700 0.2 0.2 Table B.3.: Model parameters for 6:2 FTSA sorption reactions for each layer of the 1D-vertical model and groundwater flow models Layers Fract. KL (g/m3) Nu (g/m3) Alpha (d-1) 1D-Hydrus Layer 1 0.5 2.2x10-5 0.07 0.45 1D-Hydrus Layer 2 0.5 1.4x10-5 0.07 0.15 1D-Hydrus Layer 3 0.5 1.1x10-5 0.07 0.1 1D-Hydrus Layer 4 0.5 1.0x10-8 0.07 0.1 1D-Hydrus Layer 5 0.5 1.0x10-8 0.07 0.1 1D-Hydrus Layer 6 0.5 1.0x10-8 0.07 0.1 MODFLOW/MT3DMS Layer 1 below infiltration square - 1.0x10-8 0.07 - MODFLOW/MT3DMS Layers 2-10 below infiltration square - 1.0x10-8 0.07 - MODFLOW/MT3DMS Layer 1 outside infiltration square - 1.0x10-8 0.07 - MODFLOW/MT3DMS Layers 2-10 outside infiltration square () - 1.0x10-8 0.07 - D2.4 – Guidance document on fate, transport and exposure for PMT’s 133 B. PluriMetric Pilot experiments Table B.4.: Model parameters for 6:2 FTAB sorption reactions for each layer of the 1D-vertical model and groundwater flow models Layers Fract. KL (g/m3) Nu (g/m3) Alpha (d-1) 1D-Hydrus Layer 1 0.5 5.0x10-5 0.07 0.45 1D-Hydrus Layer 2 0.5 1.9x10-5 0.07 0.25 1D-Hydrus Layer 3 0.5 1.1x10-5 0.07 0.2 1D-Hydrus Layer 4 0.5 1.0x10-7 0.07 0.2 1D-Hydrus Layer 5 0.5 1.0x10-7 0.07 0.2 1D-Hydrus Layer 6 0.5 1.0x10-7 0.07 0.2 MODFLOW/MT3DMS Layer 1 below infiltration square - 1.0x10-7 0.07 - MODFLOW/MT3DMS Layers 2-10 below infiltration square - 1.0x10-7 0.07 - MODFLOW/MT3DMS Layer 1 outside infiltration square - 1.0x10-7 0.07 - MODFLOW/MT3DMS Layers 2-10 outside infiltration square - 1.0x10-7 0.07 - Table B.5.: Model parameters for 6:2 FTSAam sorption reactions for each layer of the 1Dvertical model and groundwater flow models Layers Fract. KL (g/m3) Nu (g/m3) Alpha (d-1) 1D-Hydrus Layer 1 0.5 1.0x10-4 0.07 0.85 1D-Hydrus Layer 2 0.5 9.5x10-5 0.07 0.45 1D-Hydrus Layer 3 0.5 5.1010-5 0.07 0.4 1D-Hydrus Layer 4 0.5 1.0x10-6 0.07 0.4 1D-Hydrus Layer 5 0.5 1.0x10-6 0.07 0.4 1D-Hydrus Layer 6 0.5 1.0x10-6 0.07 0.4 D2.4 – Guidance document on fate, transport and exposure for PMT’s 134 B. PluriMetric Pilot experiments Layers (thickness) Fract. KL (g/m3) Nu (g/m3) Alpha (d-1) MODFLOW/MT3DMS Layer 1 below infiltration square - 1.0x10-6 0.07 - MODFLOW/MT3DMS Layers 2-10 below infiltration square - 1.0x10-6 0.07 - MODFLOW/MT3DMS Layer 1 outside infiltration square - 1.0x10-6 0.07 - MODFLOW/MT3DMS Layers 2-10 outside infiltration square - 1.0x10-6 0.07 - D2.4 – Guidance document on fate, transport and exposure for PMT’s 135 B. PluriMetric Pilot experiments B.4. Simulations of the spatial and temporal pattern of the non-reactive tracer concentration during the PMP experiment Figure B.1.: Measured and simulated concentrations in bromide (grey) and total organic fluoride (orange) over time in pore solution at five depths (0.03 m, 0.5 m, 1 m, 1.5 m, 2.0 m and 2.5 m depth) in the unsaturated and saturated zones below the infiltration square during the PMP experiment. Concentrations are expressed in mgF/L. Lines represent simulated concentrations over time while dots represent measured concentrations. D2.4 – Guidance document on fate, transport and exposure for PMT’s 136 References Figure B.2.: Measured and simulated concentrations in bromide (grey) and total organic fluoride (orange) over time in pore solution at three depths (1.5 m, 2.0 m and 2.5 m depth) in the saturated zone in two locations (1.2 m and 2.4 m far from the centre of the infiltration square) downstream to the infiltration square in the main flow path during the PMP experiment. Concentrations are expressed in mgF/L. Lines represent simulated concentrations over time while dots represent measured concentrations. References Bolan, N., B. Sarkar, M. Vithanage, G. Singh, D. C. W. Tsang, R. Mukhopadhyay, K. Ramadass, A. Vinu, et al. (2021). “Distribution, behaviour, bioavailability and remediation of polyand per-fluoroalkyl substances (PFAS) in solid biowastes and biowaste-treated soil”. In: Environment International 155, p. 106600. issn: 0160-4120. doi: 10.1016/j.envint.2021.106600. (Visited on 01/15/2025) (cit. on p. 127). D2.4 – Guidance document on fate, transport and exposure for PMT’s 137 C. Large scale investigations in CS#7 AFFF-polluted aquifer worst-case scenario, and that not in all training events AFFF foams were used. Therefore, it is likely that PFOS release to the soil was smaller. Figure C.4.: Vertical distribution of main HYDRUS 1D modelled variables in the unsaturated profile: a) Head in cm and distribution of materials. b) Evolution of PFOS concentrations in liquid phase. c) Lines represent modelled sorbed concentration over time for each layer, while black dots represent measured PFOS concentrations in soil at the field site. C.4. Results from fully saturated model Leachate fluxes from the unsaturated zone model were introduced in the fully saturated model at the cell representing the horizontal location of the pollution source. The unsaturated zone model was also run without pollution events, just with the rain and evapotranspiration data, to estimate effective recharge for the rest of the active top boundary cells of the saturated model. Time-dependent head data was not available for the site before remediation took place. Moreover, there are no observation piezometers from the local water agency within a reasonable distance to the site. Therefore, hydraulic permeabilities measured in the field during the characterization phase were used to parameterize the model. Concentrations measured during autumn 2021, which also do not provide information about temporal evolution of PFOS, were used to roughly calibrate transport parameters aiming to reproduce the downgradient spatial distribution of PFOS (Figure C.6). To compare observed and modelled data, modelled concentrations were vertically averaged, assuming that the field samples are a mix of the water entering the screened interval of the monitoring well. D2.4 – Guidance document on fate, transport and exposure for PMT’s 144 C. Large scale investigations in CS#7 AFFF-polluted aquifer Figure C.5.: Fluxes and heads at top and bottom boundaries. a) Precipitation from Fogars de la Selva meteo station plus the estimated infiltration during each firefighting training event. The potential evapotranspiration (ETP) estimated with Hargreaves. b) Fluxes through top (vTop-blue) and bottom (vBot - green) boundaries. c) Head at top (hTop) and bottom (hBot) boundaries. d) Concentration at top boundary (cTop – left axis). Concentration imposed at the atmospheric boundary condition (Conc) and concentration existing the bottom boundary (cBot). D2.4 – Guidance document on fate, transport and exposure for PMT’s 145 C. Large scale investigations in CS#7 AFFF-polluted aquifer Figure C.6.: Downgradient (from the source) distribution of most abundant PFAS at the site. Data obtained during a sampling campaign in autumn 2021. Dashed line represents modelled concentrations of PFOS. Modelled PFOS concentrations in groundwater (Figure C.6) are in the order of magnitude of observed ones, though still higher. Oscillations in the concentration of the recharge flux at the source are reflected as higher concentration pulses in the downflowing plume (Figure C.7). They can also be observed as little peaks downflow from the source in Figure C.6. Time zero represents the moment in which PFOS are release for the first time at the surface. It takes them around a year to reach the groundwater level. From there, it takes around 2 years to reach the river (Figure C.8). C.5. Conclusions The modelling approach presented here combines HYDRUS1D and MODFLOW/MT3D to model PFAS fate and transport from the source at the soil surface, to the exfiltration point, in a downgradient river. The model uses a simple linear sorption to approximate observed PFOS concentration in soil and liquid phases. Based on the model a rough estimation of the time for the PFOS plume to reach the river was obtained. The presented modelling approach simplifies the behaviour of PFAS. In reality, sorption at the soil matrix does not have a linear isotherm. Moreover, sorption at the air-water interface, which is very relevant for some PFAS, is not incorporated in the Hydrus-1D code. Therefore, this sorption process can only be incorporated by la constant sorption isotherm. As long as the ratio of the soil-water content to the area of the water air interface remains constant, this approximation is acceptable. Moreover, any more complex isotherm will need better data to estimate the isotherm parameters. The complexity of the model should go in line with the D2.4 – Guidance document on fate, transport and exposure for PMT’s 146 C. Large scale investigations in CS#7 AFFF-polluted aquifer Figure C.7.: PFOS concentration distribution at 6 modelled times (days). Concentrations in g/m3. D2.4 – Guidance document on fate, transport and exposure for PMT’s 147 References Figure C.8.: Evolution of concentrations of PFOS in groundwater at different depth for the two boreholes close to the source and for the river amount and quality of the available data. In the absence of lots of data, the combination of HYDRUS1D and MODFLOW/MT3D, through their Phyton tools, could be a good starting point for modelling the fate and transport of PFAS. In an optimal situation, in which sufficient spatial and temporal coverage of the data is available, the main limitations of this approach are, (a) the lack of a real coupling between both models, and (b) the lack of the incorporation of the process of adsorption of PFAS at the air-water interface in Hydrus-1D. The effect of temporal changes in the water content on the ratio of adsorbed and dissolved PFAS is therefore not described correctly. This assumes the fringe zone does not move, neglecting the contribution of draining and rewatering cycles in the transport of PFOS towards the groundwater. This is an important process for PFAS transport, as they are substances that concentrate at the interface between air and water. References Collenteur, R., G. Brunetti, and M. Vremec (2019). Phydrus: Python implementation of the HYDRUS-1D unsaturated zone model (cit. on p. 140). Goldenman, G., M. Fernandes, M. Holland, T. Tugran, A. Nordin, C. Schoumacher, and A. McNeill (Mar. 2019). The cost of inaction. en. 2019:516. TemaNord. Copenhagen: Nordic Council of Ministers. isbn: 978-92-893-6065-4. doi: 10.6027/TN2019-516. url: http://urn.kb. se/resolve?urn=urn:nbn:se:norden:org:diva-5514 (visited on 01/28/2025) (cit. on p. 140). Held, T. and M. Reinhard (2020). Remediation management for local and wide-spread PFAS contaminations. en. Umweltbundesamt. url: https://www.umweltbundesamt.de/en/ publikationen/remediation-management-for-local-wide-spread-pfas (visited on 01/15/2025) (cit. on p. 140). D2.4 – Guidance document on fate, transport and exposure for PMT’s 148 References Høisæter, Å., A. Pfaff, and G. D. Breedveld (2019). “Leaching and transport of PFAS from aqueous film-forming foam (AFFF) in the unsaturated soil at a firefighting training facility under cold climatic conditions”. In: Journal of Contaminant Hydrology 222, pp. 112–122. issn: 0169-7722. doi: 10.1016/j.jconhyd.2019.02.010 (cit. on pp. 142, 143). Hughes, J. D., C. D. Langevin, S. R. Paulinski, J. D. Larsen, and D. Brakenhoff (2023). “FloPy Workflows for Creating Structured and Unstructured MODFLOW Models”. en. In: Groundwater 62.1, pp. 124–139. issn: 1745-6584. doi: 10.1111/gwat.13327 (cit. on p. 142). López de Alda, M., M. Llorca, M. Farré, A. Cano López, V. Matamoros Mercadal, L. Fernández Rojo, and C. Bosch (2025). PROMISCES, D2.2 – Characterization of PFAS and chlorinated solvent contamination in two aquifers in Spain. in preparation. doi: 10.3030/101036449. url: https://cordis.europa.eu/project/id/101036449/results (cit. on p. 140). McGarr, J. T., E. G. Mbonimpa, D. C. McAvoy, and M. R. Soltanian (May 2023). “Fate and Transport of Perand Polyfluoroalkyl Substances (PFAS) at Aqueous Film Forming Foam (AFFF) Discharge Sites: A Review”. en. In: Soil Systems 7.2, p. 53. issn: 2571-8789. doi: 10.3390/soilsystems7020053 (cit. on p. 143). D2.4 – Guidance document on fate, transport and exposure for PMT’s 149 D. Catchment Model MoRE for the upper Danube basin Authored by: Meiqi Liu1, Steffen Kittlaus1, Matthias Zessner1 1TU Wien, Institute of Water Quality and Resource Management, Karlsplatz 13, Vienna, Austria D.1. Overview This Annex provides an overview of the regionalized emission model (MoRE) developed as part of Case Study 2 in the PROMISCES project, along with the final results generated from this model. We applied the model to a spatial scale that covers the Upper Danube basin, from its source to the city of Budapest. Figure D.1.: Overview map showing the Danube catchment (outlined in red) and the case study #2 area–Upper Danube catchment (outlined in black), including smaller catchment units. D2.4 – Guidance document on fate, transport and exposure for PMT’s 150 D. Catchment Model MoRE for the upper Danube basin D.2. Modelled PFASs In PROMISCES Case Study 2, the focus of the MoRE model is on perand polyfluoroalkyl substances (PFASs), a group of synthetic chemicals widely used in various household and industrial applications (Glüge et al. 2020). Based on the results of the monitoring campaign conducted under Subtask 2.2.4 – Large catchment scale monitoring, we selected a total number of 18 PFAS substances as input to the model. In addition, surface water monitoring data from selected river gauges provided the opportunity to validate the model. Table D.1.: List of modelled PFAS substances. The substance groups are classified according to the EPA Method 1633 (EPA 2024) Short Name Substances CAS No. Substance Group PFBA Perfluorobutanoic acid 375-22-4 short-chain Perfluoroalkyl carboxylic acids PFPeA Perfluoropentanoic acid 2706-90-3 short-chain Perfluoroalkyl carboxylic acids PFHxA Perfluorohexanoic acid 307-24-4 short-chain Perfluoroalkyl carboxylic acids PFHpA Perfluoroheptanoic acid 375-85-9 short-chain Perfluoroalkyl carboxylic acids PFOA Perfluoroctanoic acid 335-67-1 long-chain Perfluoroalkyl carboxylic acids PFNA Perfluorononaoic acid 375-95-1 long-chain Perfluoroalkyl carboxylic acids PFDA Perfluorodecanoic acid 335-76-2 long-chain Perfluoroalkyl carboxylic acids PFBS Perfluorobutane sulfonic acid 375-73-5 short-chain Perfluoroalkyl sulfonic acids PFPeS Perfluoropentane sulfonic acid 2706-91-4 short-chain Perfluoroalkyl sulfonic acids D2.4 – Guidance document on fate, transport and exposure for PMT’s 151 D. Catchment Model MoRE for the upper Danube basin Short Name Substances CAS No. Substance Group PFHxS Perfluorohexane sulfonic acid 355-46-4 short-chain Perfluoroalkyl sulfonic acids PFOS Perfluoroctane sulfonic acid 1763-23-1 long-chain Perfluoroalkyl sulfonic acids PFNS Perfluorononane sulfonic acid 68259-12-1 long-chain Perfluoroalkyl sulfonic acids PFDS Perfluorodecane sulfonic acid 335-77-3 long-chain Perfluoroalkyl sulfonic acids FTSA62 6:2 Fluorotelomer sulfonic acid 27619-97-2 Fluorotelomer sulfonic acids ADONA 4,8-Dioxa-3H-perfluorononanoic acid 919005-14-4 Perand Polyfluoroether carboxylic acids GenX Hexafluoropropylene oxide dimer acid 13252-13-6 Perand Polyfluoroether carboxylic acids PFOSA Perfluorooctanesulfonamide 754-91-6 Perfluorooctane sulfonamides NEtFOSAA N-ethyl perfluorooctanesulfonamidoacetic acid 2991-50-6 Perfluorooctane sulfonamidoacetic acids D.3. Modelled Pathways The MoRE model system was originally designed as a pathway-oriented tool to simulate the input of substances into surface waters (Fuchs et al. 2017). In the PROMISCES project, we adapted the MoRE model to include PFASs and identified potential critical pathways for PFAS emissions, following the guidelines of the EU Guidance Document No.28 (European Comission (EC) 2012). For general details, please refer to section 4.2 of the main text of the guidance document. Overall, the model includes eight pathways, with the groundwater (P4), urban sewer system (P7) and industrial direct emitter (P10) pathways further divided into several sub-pathways. In addition to the base model and scenarios built on its setup, we also implemented a version of the model that accounts for groundwater diffuse pollution from landfills and aerodromes that might conduct fire-fighting activities. Results related to this version are not included in this document, but available in the exported SQLite model. For further information, please refer to the PROMISCS Toolbox document (Groot et al. 2025). D2.4 – Guidance document on fate, transport and exposure for PMT’s 152 D. Catchment Model MoRE for the upper Danube basin Table D.2.: List of pathways with abbreviations and corresponding numbers as referenced in the European Comission (EC) (2012) document, modelled in this study. Number Abbreviation Description of the pathway P1 AD Atmospheric deposition directly to surface water P2 ER Erosion P3 SR Surface runoff from unsealed areas P4a GW Groundwater background P4b LS Groundwater legacy pollution from the Gendorf site P4c DU Inhabitant-specific diffuse pollution via groundwater P7a US_ss Strom sewers in the separate sewer system P7b US_cso Combined sewer overflows P7c US_oss Sewer connected but not going to WWTP P8 WWTP Urban wastewater treatment plants P9 US_nss Individual - treated and untreated - household discharge P10a ID Industrial wastewater treated P10b ID_Gendorf Direct wastewater discharge from the industrial park Gendorf D.4. Basic Model Setup The model encompasses 526 sub-catchments, with an average area of 354 ±352 km2, and operates on an annual temporal resolution. The base model covers the period from 2015 to 2021, which is defined as the "reference period." For scenario setups, the years 2013 and 2018 were selected to represent baseline years for preand post-climate change situations, respectively. For detailed setup on scenarios, please refer to the Annex of this guidance document on Model Scenarios. Hydrological input parameters for the model were derived from the Wflow model developed by Deltares (Verseveld et al. 2024), with enhanced resolution achieved by combing precipitation data from the scaled SPARTACUS dataset GeoSphere Austria (2020) and the ERA5 dataset (Hersbach et al. 2023). Other basic input data were sourced from various open databases or directly requested from local authorities. PFAS-specific input data were collected from the Subtask 2.2.4 monitoring activities, and from the DHm3c inventory (Kittlaus et al. 2023). Since many concentration data points were below the limit of quantifications (LOQ), the regression on order statistics (ROS) (Helsel 2012) method was applied to help more accurate estimation on summary statistics. When the data met the requirements for the ROS method (more than 3 detects and more than 20% detected), D2.4 – Guidance document on fate, transport and exposure for PMT’s 153