the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Feb 2025
Explicit simulation of reactive microbial transport with a dual-permeability, two-site kinetic deposition formulation using the integrated surface-subsurface hydrological model HydroGeoSphere
Abstract. Assessing the transport behavior of microbes in surface water-groundwater systems is important to prevent contamination of drinking water resources by pathogens. While wellhead protection area (WHPA) delineation is still predominantly based on dye injection tests and advective transport modeling, size exclusion of colloid-sized microbes from the smaller and usually less conductive pore space causes a faster breakthrough and thus faster apparent transport of microbes compared to that of solutes. To provide a tool for better assessment of the differences between solute and microbial transport in surface water-groundwater systems, we here present the implementation of a dual-permeability, two-site kinetic deposition formulation for microbial transport in the integrated surface-subsurface hydrological model HydroGeoSphere (HGS). The implementation considers attachment, detachment and inactivation of microbes in both permeability regions and allows for multispecies transport. The dual-permeability, two-site kinetic deposition implementation in HGS was verified against an analytical solution for dual-permeability colloid transport and the suitability of the model for microbial transport at the wellfield scale is illustrated in a multi-tracer flow and transport study of an idealized alluvial riverbank filtration site. In this illustrative example, the transport of reactive microbes, conservative 4He, and reactive 222Rn was simulated in parallel, allowing mixing ratios, tracer breakthrough curves and travel times to be assessed via multiple approaches. The developed simulation tool is the first integrated surface-subsurface hydrological simulator for reactive solute and microbial transport, and marks an important advancement to unlock and quantify governing microbial transport processes in riverbank filtration settings. It enables meaningful WHPA delineation and risk assessments even under extreme hydrological situations such as flood events.
- Preprint
(1337 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2025-372', Anonymous Referee #1, 08 Mar 2025
This manuscript presents the implementation of an existing dual-permeability approach to address transport of (microbial) colloids in the proprietary software HydroGeoSphere that can simulate saturated-unsaturated flow and transport coupled to surface flow and transport. The code is validated against an existing analytical solution for 1-D dual-permeability transport with linear exchange kinetics between the two domains and linear kinetic sorption in both of them. The final demonstration is a virtual test case mimicking bank filtration, in which colloidal transport is simulated together with 4He and radon as natural tracers. There is no comparison to experimental or field data.
As the manuscript’s major statement is that the authors have transferred an existing model formulation to an existing software package I see this as a technical note rather than a research article. The 1-D example is not particularly exciting from a process-insight view because the influence of different parameters on multi-permeability models has been discussed before. I have remarks on the 3-D application further down.
With respect to presenting the model extension of HydroGeoSphere, the text is written in a slightly odd way. It is not clear what was actually developed within the study, and what has already existed. I checked the HydroGeoSphere documentation, which includes colloid transport in dual-permeability models. But that might only mean that the documentation has already included the results of the present study. The only wording that directly implies model extension is in lines 175-177, where the inclusion of a first-order decay term is mentioned. Please clarify that the implementation of the Bradford et al. (2009, doi: 10.1029/2008WR007096) formulation is really specific to this study. Otherwise I am even more puzzled what the message of the manuscript is.
Let’s come to the conceptual assumptions of the Bradford et al. (2009) formulation. Dual-permeability flow and transport was introduced by Barenblatt et al. (1960, doi: 10.1016/0021-8928(60)90107-6) to address preferential flow in fractured media, and has been used to parameterize effects on heterogeneity on (flow and) solute transport, characterized by strong anomalous behavior. The concept has been presented using different names (e.g., multi-domain model, mobile-mobile model). In contrast to dual-porosity (mobile-immobile, transient storage) models, it requires solving two coupled flow problems, posing big difficulties in unique calibration if the flow behavior is quite normal. Bradford et al. (2009) reinterpreted the conceptual model to facilitate that colloids break through earlier than solutes. The latter effect is caused by size exclusion and by the fact that colloids cannot experience the velocity within pores at distances to solids smaller than half their diameter. If the colloids have the same net electrical charge as the grain surface they experience an even smaller portion of the intra-pore velocity distribution because they are repelled from the no-slip boundary. Other model formulations for colloid transport, involving reversible attachment-detachment and irreversible straining, cannot reproduce a first breakthrough before that of solutes. Bradford et al. (2009) claimed that their formulation also addresses straining, but that is not really true. In both domains, the model assumes kinetic first-order attachment and detachment. It is trivial to derive equilibrium sorption coefficients from the ratio of the detachment and attachment rate coefficients. Choosing a low detachment coefficient in the low-permeability domain still implies reversible sorption. However, straining is irreversible: particles get stuck in pore throats and never ever become mobile again. This implies that the formulation of Bradford et al. (2009) leads to very long tailing of colloid breakthrough curves with a complete recovery at infinite times, whereas models that include an elimination mechanism according to standard colloid filtration theory will have an incomplete recovery (i.e. the zeroth moment of the transfer function between the in- and output concentration in 1-D transport is smaller than unity). If you really want straining, you either choose a detachment coefficient in the less-mobile domain of zero, or you introduce a first-order elimination term. The authors have such a term, but they relate it other mechanisms than straining because they uncritically adopt the erroneous perspective of Bradford et al. (2009) that kinetic reversible mass transfer could parameterize (irreversible) straining.
Would there be a much simpler way of achieving an earlier breakthrough of colloids than of solutes without introduction of a second permeability? That is indeed possible. All you need is a retardation factor for the colloids smaller than one, and they are faster than the solutes. You would still need an irreversible straining term and (at least one) kinetic, reversible attachment-detachment term, but the model would be much simpler because you could skip coming up with two spatial permeability distributions and exchange terms of the fluids between the two domains (and the flow problem would have half the number of unknowns). Given the fact that the authors don’t want to use the dual-permeability formulation for its original purpose (addressing anomalous flow and transport in highly heterogeneous formations), and that they don’t have the data to inform such a model for its original purpose, I cannot recommend the conceptual approach chosen by the authors. It is simply an overkill, particularly in 3-D settings.
The authors promise a reactive microbial transport model in the title. This is slightly misleading. The only “reaction” term is a first-order elimination term. Many microbes of interest undergo complex dynamics due to growth, dormancy and reactivation, change in mobility corresponding to their physiological state or abundance. In the model of the authors the microbes must be introduced via the inflow and can only adsorb or vanish besides transport. They are essentially treated like dead particles. I can fully understand that the authors don’t intend to elaborate on microbial dynamics, but then they should be careful in selling they model as “reactive microbial transport model”.
As mentioned above, I am not too impressed by the 1-D tests. They mainly reproduce the work of Bradford et al. (2009), who at least had a comparison to real measurements, and of Leij and Bradford (2013). It is of course important that a code is tested against analytical solutions for model validation, but it is not a scientifically particularly exciting exercise.
The 3-D demonstration is supposed to show that the model works field-similar settings. It is not well suited to convince the reader that an integrated surface-subsurface hydrological model is needed. The test would equally well work with a pure groundwater model forced by the boundary condition at the river. In the setup, like in many bank filtration applications, the river is not really affected by groundwater flow and transport (and every HydroGeoSphere user knows that running the model as pure porous-medium model makes life much easier). A more interested application would show real feedbacks between the surface and subsurface domains, e.g., when considering transport of pathogens in a meandering stream with intensive hyporheic exchange and bank storage. That is obviously not the final application that the authors have in mind, but I would claim that the 3-D test problem could be simulated with loose coupling of the surface and subsurface domains (or even just predefining river-stage and concentration fluctuations as boundary condition).
In the 3-D test case, the authors first simulate steady-state concentrations, involving an input of (microbial) colloids from the river, 4He as a natural tracer for the mixing of old groundwater with river water, and 222Rn as age tracer. They then simulate the response to a river-stage fluctuation. Proportional to the river discharge they increase the concentration of the (microbial) colloids an of 4He in the river (mimicking an artificial tracer with concentration is proportional to that of the microbes). The former makes sense, whereas I am not happy about the later because 4He is supposed to be an indicator of mixing of old groundwater and river infiltrate, and with the pulse in the river the boundary between the two water bodies moves. While this boundary may be far away from the observation points, it is not particularly smart to create two causes of 4He changes. It might have been better to add a real artificial tracer to separate the signals.
As designed, the colloids break slightly earlier through than the solute tracer. The effect is not super big (4.5% earlier peak time) and well within the uncertainty of travel-time estimates in real-world studies on outlying well-head protection zones. With a longitudinal dispersivity of more than 5m (and additional dispersion caused by mobile-mobile transport), the peaks are so broad that the difference in the breakthrough curves are not particularly obvious by eye sight. The much more interesting signal is that of the radon. Here, the authors get a much earlier breakthrough. They attribute this to radon following the pressure wave (lines 433-434), but that makes physically no sense. What I believe is that the river-stage fluctuation shifts the flow pattern and by that the age distribution. Honestly, I find this phenomenon more interesting than the micobial-transport study as you see a real signal.
The results section ends with a summary/conclusion, followed by a discussion section, and then final conclusions. That’s a little bit odd, particularly since the actual conclusions are quite shallow.
In summary, I have expressed my doubts that the dual-permeability model is the best choice for transport of microbial colloids. I am convinced that you can achieve the same results computationally much cheaper. I believe that the 1-D model has too much weight given that it includes nothing new. The 3-D application does not need an integrated surface-subsurface model and does not underscore that the chosen model formulation is really needed. If there were real data that can only be interpreted with the model, the authors would have a much stronger point. This manuscript needs severe revisions to make it a significant contribution.
Citation: https://doi.org/10.5194/egusphere-2025-372-RC1 -
AC1: 'Reply on RC1', Friederike Currle, 02 Apr 2025
Dear reviewers, dear editors
We are pleased with the feedback and would like to thank the two reviewers for the very helpful comments. None of the reviewers have risen any technical concerns related to the implementation of the code, which highlights that our code is relevant and carefully implemented. However, both reviewers have suggested some important improvements in the presentation of the verification and illustration examples of the new model implementation, which we will implement, and which will allow the manuscript to improve further. Moreover, as correctly pointed out by reviewer #1, the paper is a technical note rather than a research paper, an aspect which we would like to confirm.
We will address all comments of reviewer #1 in a revised manuscript as outlined in detail in the supplement.
Sincerely,
Friederike Currle, René Therrien and Oliver Schilling
-
AC1: 'Reply on RC1', Friederike Currle, 02 Apr 2025
-
RC2: 'Comment on egusphere-2025-372', Anonymous Referee #2, 17 Mar 2025
This manuscript outlines the inclusion of dual-permeability, two-site kinetic deposition formula for microbial transport in HydroGeoSphere (HGS). I believe this topic is relevant and of interest to the readers of EGUsphere. The manuscript is generally well-written, with the inclusions of expected sections outlining the model development, validation and illustrative application, all written in a clear manner.
As presented, my primary concern relates to the novelty and contributions of this work. How does this differentiate from other subsurface reactive transport models available – why should someone choose to use HGS over these models? I understand that the primary feature of including these equations in HGS is the inclusion with an integrated hydrologic model as opposed to solely a subsurface model, but no part of this manuscript uses or highlights the benefit of an integrated approach over subsurface-only. The authors outline the equations that govern surface flow and transport, but the verification and illustrative application do not seem to utilize the surface domain at all. Perhaps it is simulated, but the results are not presented or discussed. I think this is critical to the contributions of this work – what does this newly developed feature provide that was otherwise lacking? Perhaps a comparison between groundwater-only simulations and the integrated approaches can help demonstrate the benefit of including these equations with HGS as opposed to solely a subsurface approach.
Given this concern is critical to the contributions and novelty of this research, I feel the authors need to make significant revisions before this manuscript can be considered for publication.
Citation: https://doi.org/10.5194/egusphere-2025-372-RC2 -
AC2: 'Reply on RC2', Friederike Currle, 02 Apr 2025
Dear reviewers, dear editors
We are pleased with the feedback and would like to thank the two reviewers for the very helpful comments. None of the reviewers have risen any technical concerns related to the implementation of the code, which highlights that our code is relevant and carefully implemented. However, both reviewers have suggested some important improvements in the presentation of the verification and illustration examples of the new model implementation, which we will implement, and which will allow the manuscript to improve further. Moreover, as correctly pointed out by reviewer #1, the paper is a technical note rather than a research paper, an aspect which we would like to confirm.
We will address all comments of reviewer #2 in a revised manuscript as outlined in detail in the supplement.
Sincerely,
Friederike Currle, René Therrien and Oliver Schilling
-
AC2: 'Reply on RC2', Friederike Currle, 02 Apr 2025
Status: closed
-
RC1: 'Comment on egusphere-2025-372', Anonymous Referee #1, 08 Mar 2025
This manuscript presents the implementation of an existing dual-permeability approach to address transport of (microbial) colloids in the proprietary software HydroGeoSphere that can simulate saturated-unsaturated flow and transport coupled to surface flow and transport. The code is validated against an existing analytical solution for 1-D dual-permeability transport with linear exchange kinetics between the two domains and linear kinetic sorption in both of them. The final demonstration is a virtual test case mimicking bank filtration, in which colloidal transport is simulated together with 4He and radon as natural tracers. There is no comparison to experimental or field data.
As the manuscript’s major statement is that the authors have transferred an existing model formulation to an existing software package I see this as a technical note rather than a research article. The 1-D example is not particularly exciting from a process-insight view because the influence of different parameters on multi-permeability models has been discussed before. I have remarks on the 3-D application further down.
With respect to presenting the model extension of HydroGeoSphere, the text is written in a slightly odd way. It is not clear what was actually developed within the study, and what has already existed. I checked the HydroGeoSphere documentation, which includes colloid transport in dual-permeability models. But that might only mean that the documentation has already included the results of the present study. The only wording that directly implies model extension is in lines 175-177, where the inclusion of a first-order decay term is mentioned. Please clarify that the implementation of the Bradford et al. (2009, doi: 10.1029/2008WR007096) formulation is really specific to this study. Otherwise I am even more puzzled what the message of the manuscript is.
Let’s come to the conceptual assumptions of the Bradford et al. (2009) formulation. Dual-permeability flow and transport was introduced by Barenblatt et al. (1960, doi: 10.1016/0021-8928(60)90107-6) to address preferential flow in fractured media, and has been used to parameterize effects on heterogeneity on (flow and) solute transport, characterized by strong anomalous behavior. The concept has been presented using different names (e.g., multi-domain model, mobile-mobile model). In contrast to dual-porosity (mobile-immobile, transient storage) models, it requires solving two coupled flow problems, posing big difficulties in unique calibration if the flow behavior is quite normal. Bradford et al. (2009) reinterpreted the conceptual model to facilitate that colloids break through earlier than solutes. The latter effect is caused by size exclusion and by the fact that colloids cannot experience the velocity within pores at distances to solids smaller than half their diameter. If the colloids have the same net electrical charge as the grain surface they experience an even smaller portion of the intra-pore velocity distribution because they are repelled from the no-slip boundary. Other model formulations for colloid transport, involving reversible attachment-detachment and irreversible straining, cannot reproduce a first breakthrough before that of solutes. Bradford et al. (2009) claimed that their formulation also addresses straining, but that is not really true. In both domains, the model assumes kinetic first-order attachment and detachment. It is trivial to derive equilibrium sorption coefficients from the ratio of the detachment and attachment rate coefficients. Choosing a low detachment coefficient in the low-permeability domain still implies reversible sorption. However, straining is irreversible: particles get stuck in pore throats and never ever become mobile again. This implies that the formulation of Bradford et al. (2009) leads to very long tailing of colloid breakthrough curves with a complete recovery at infinite times, whereas models that include an elimination mechanism according to standard colloid filtration theory will have an incomplete recovery (i.e. the zeroth moment of the transfer function between the in- and output concentration in 1-D transport is smaller than unity). If you really want straining, you either choose a detachment coefficient in the less-mobile domain of zero, or you introduce a first-order elimination term. The authors have such a term, but they relate it other mechanisms than straining because they uncritically adopt the erroneous perspective of Bradford et al. (2009) that kinetic reversible mass transfer could parameterize (irreversible) straining.
Would there be a much simpler way of achieving an earlier breakthrough of colloids than of solutes without introduction of a second permeability? That is indeed possible. All you need is a retardation factor for the colloids smaller than one, and they are faster than the solutes. You would still need an irreversible straining term and (at least one) kinetic, reversible attachment-detachment term, but the model would be much simpler because you could skip coming up with two spatial permeability distributions and exchange terms of the fluids between the two domains (and the flow problem would have half the number of unknowns). Given the fact that the authors don’t want to use the dual-permeability formulation for its original purpose (addressing anomalous flow and transport in highly heterogeneous formations), and that they don’t have the data to inform such a model for its original purpose, I cannot recommend the conceptual approach chosen by the authors. It is simply an overkill, particularly in 3-D settings.
The authors promise a reactive microbial transport model in the title. This is slightly misleading. The only “reaction” term is a first-order elimination term. Many microbes of interest undergo complex dynamics due to growth, dormancy and reactivation, change in mobility corresponding to their physiological state or abundance. In the model of the authors the microbes must be introduced via the inflow and can only adsorb or vanish besides transport. They are essentially treated like dead particles. I can fully understand that the authors don’t intend to elaborate on microbial dynamics, but then they should be careful in selling they model as “reactive microbial transport model”.
As mentioned above, I am not too impressed by the 1-D tests. They mainly reproduce the work of Bradford et al. (2009), who at least had a comparison to real measurements, and of Leij and Bradford (2013). It is of course important that a code is tested against analytical solutions for model validation, but it is not a scientifically particularly exciting exercise.
The 3-D demonstration is supposed to show that the model works field-similar settings. It is not well suited to convince the reader that an integrated surface-subsurface hydrological model is needed. The test would equally well work with a pure groundwater model forced by the boundary condition at the river. In the setup, like in many bank filtration applications, the river is not really affected by groundwater flow and transport (and every HydroGeoSphere user knows that running the model as pure porous-medium model makes life much easier). A more interested application would show real feedbacks between the surface and subsurface domains, e.g., when considering transport of pathogens in a meandering stream with intensive hyporheic exchange and bank storage. That is obviously not the final application that the authors have in mind, but I would claim that the 3-D test problem could be simulated with loose coupling of the surface and subsurface domains (or even just predefining river-stage and concentration fluctuations as boundary condition).
In the 3-D test case, the authors first simulate steady-state concentrations, involving an input of (microbial) colloids from the river, 4He as a natural tracer for the mixing of old groundwater with river water, and 222Rn as age tracer. They then simulate the response to a river-stage fluctuation. Proportional to the river discharge they increase the concentration of the (microbial) colloids an of 4He in the river (mimicking an artificial tracer with concentration is proportional to that of the microbes). The former makes sense, whereas I am not happy about the later because 4He is supposed to be an indicator of mixing of old groundwater and river infiltrate, and with the pulse in the river the boundary between the two water bodies moves. While this boundary may be far away from the observation points, it is not particularly smart to create two causes of 4He changes. It might have been better to add a real artificial tracer to separate the signals.
As designed, the colloids break slightly earlier through than the solute tracer. The effect is not super big (4.5% earlier peak time) and well within the uncertainty of travel-time estimates in real-world studies on outlying well-head protection zones. With a longitudinal dispersivity of more than 5m (and additional dispersion caused by mobile-mobile transport), the peaks are so broad that the difference in the breakthrough curves are not particularly obvious by eye sight. The much more interesting signal is that of the radon. Here, the authors get a much earlier breakthrough. They attribute this to radon following the pressure wave (lines 433-434), but that makes physically no sense. What I believe is that the river-stage fluctuation shifts the flow pattern and by that the age distribution. Honestly, I find this phenomenon more interesting than the micobial-transport study as you see a real signal.
The results section ends with a summary/conclusion, followed by a discussion section, and then final conclusions. That’s a little bit odd, particularly since the actual conclusions are quite shallow.
In summary, I have expressed my doubts that the dual-permeability model is the best choice for transport of microbial colloids. I am convinced that you can achieve the same results computationally much cheaper. I believe that the 1-D model has too much weight given that it includes nothing new. The 3-D application does not need an integrated surface-subsurface model and does not underscore that the chosen model formulation is really needed. If there were real data that can only be interpreted with the model, the authors would have a much stronger point. This manuscript needs severe revisions to make it a significant contribution.
Citation: https://doi.org/10.5194/egusphere-2025-372-RC1 -
AC1: 'Reply on RC1', Friederike Currle, 02 Apr 2025
Dear reviewers, dear editors
We are pleased with the feedback and would like to thank the two reviewers for the very helpful comments. None of the reviewers have risen any technical concerns related to the implementation of the code, which highlights that our code is relevant and carefully implemented. However, both reviewers have suggested some important improvements in the presentation of the verification and illustration examples of the new model implementation, which we will implement, and which will allow the manuscript to improve further. Moreover, as correctly pointed out by reviewer #1, the paper is a technical note rather than a research paper, an aspect which we would like to confirm.
We will address all comments of reviewer #1 in a revised manuscript as outlined in detail in the supplement.
Sincerely,
Friederike Currle, René Therrien and Oliver Schilling
-
AC1: 'Reply on RC1', Friederike Currle, 02 Apr 2025
-
RC2: 'Comment on egusphere-2025-372', Anonymous Referee #2, 17 Mar 2025
This manuscript outlines the inclusion of dual-permeability, two-site kinetic deposition formula for microbial transport in HydroGeoSphere (HGS). I believe this topic is relevant and of interest to the readers of EGUsphere. The manuscript is generally well-written, with the inclusions of expected sections outlining the model development, validation and illustrative application, all written in a clear manner.
As presented, my primary concern relates to the novelty and contributions of this work. How does this differentiate from other subsurface reactive transport models available – why should someone choose to use HGS over these models? I understand that the primary feature of including these equations in HGS is the inclusion with an integrated hydrologic model as opposed to solely a subsurface model, but no part of this manuscript uses or highlights the benefit of an integrated approach over subsurface-only. The authors outline the equations that govern surface flow and transport, but the verification and illustrative application do not seem to utilize the surface domain at all. Perhaps it is simulated, but the results are not presented or discussed. I think this is critical to the contributions of this work – what does this newly developed feature provide that was otherwise lacking? Perhaps a comparison between groundwater-only simulations and the integrated approaches can help demonstrate the benefit of including these equations with HGS as opposed to solely a subsurface approach.
Given this concern is critical to the contributions and novelty of this research, I feel the authors need to make significant revisions before this manuscript can be considered for publication.
Citation: https://doi.org/10.5194/egusphere-2025-372-RC2 -
AC2: 'Reply on RC2', Friederike Currle, 02 Apr 2025
Dear reviewers, dear editors
We are pleased with the feedback and would like to thank the two reviewers for the very helpful comments. None of the reviewers have risen any technical concerns related to the implementation of the code, which highlights that our code is relevant and carefully implemented. However, both reviewers have suggested some important improvements in the presentation of the verification and illustration examples of the new model implementation, which we will implement, and which will allow the manuscript to improve further. Moreover, as correctly pointed out by reviewer #1, the paper is a technical note rather than a research paper, an aspect which we would like to confirm.
We will address all comments of reviewer #2 in a revised manuscript as outlined in detail in the supplement.
Sincerely,
Friederike Currle, René Therrien and Oliver Schilling
-
AC2: 'Reply on RC2', Friederike Currle, 02 Apr 2025
Data sets
Supporting Information datasets for article: "Explicit simulation of reactive microbial transport with a dual-permeability, two-site kinetic deposition formulation using the integrated surface-subsurface hydrological model HydroGeoSphere" Friederike Currle, René Therrien, and Oliver S. Schilling https://doi.org/10.4211/hs.401dedd41b7040808482019759abc42c
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 314 | 72 | 25 | 411 | 16 | 38 |
- HTML: 314
- PDF: 72
- XML: 25
- Total: 411
- BibTeX: 16
- EndNote: 38
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1