Biomagnetic Monitoring vs. CFD Modeling: A Real Case Study of Near-Source Depositions of Traffic-Related Particulate Matter along a Motorway

Round 1

Reviewer 1 Report

In this revised article, Letaïef and coauthors describe a monitoring and modeling study of near-source deposition of traffic-related particulate matter along a motorway in France.  This revision clarified the scope of the article but it remains structured as a biomagnetic monitoring study and CFD modeling study that are linked geographically but with results that are only qualitatively comparable.  The biomagnetic monitoring effectively attributes the metallic deposition on plants, soil, and artificial filters to traffic, but is not applicable as an evaluation dataset for the concentration-based CFD modeling besides suggesting a recirculated flow behind the berm/wall.  Separately, I don't believe that the monitoring and modeling studies have enough originality to be publishable.  I also don't believe that the monitoring design as presented is capable of being used to evaluate the CFD model.  Therefore, I'd suggest that this article be rejected.  If the editor comes to a different conclusion, I'd also strongly suggest additional English language editing, particularly in the Abstract and Introduction sections of the article.

Author Response

We are sorry that our work did not convince you. You do not believe in our scientific approach without however putting forward a counter argument that would allow us to improve the quality of our manuscript. We regret this because our sincere desire is to produce the best possible study. You acknowledge that the bibliography has been correctly compiled in our work. Also, by browsing this bibliography, you may have noticed that the techniques of environmental magnetism is a technique used in dozens of studies published in scientific journals of first rank quality. This technique only makes it possible to account for the relative variation of dust deposition on accumulating surfaces. However, this unconventional technique seems to show a strong potential for detecting the most impacted areas which are located near the sources of pollutant emissions. This technique only provides qualitative information and we are perfectly aware that it should be used with the utmost care. We thought that one way to see how far we could use this technique was to compare such experimental data with a CFD outcomes. Doing thus, we clearly show in this contribution that this technique may prove to be relevant for air quality monitoring but still acknowledging that more work are required for a more precise calibration. We do not share your conclusion on the originality of the CFD model that we developed. To our knowledge, few (or no) models of dispersion of pollutants from traffic include in the atmospheric wind field the turbulence and the moment induced by the real traffic, in the present case 240,000 vehicles distributed daily on 12 carriage ways. We acknowledge that this model is perfectible but as already mentioned in the manuscript, all the source code and test cases are available on Zenodo repository with the great hope that they could be improved by the community in a near future.

Reviewer 2 Report

I have gone through the revised manuscript and the cover letter. The authors have addressed my comments appropriately.

Author Response

We are very grateful for the relevant comments and the working time spent to considerably improve our manuscript by the reviewer.

Reviewer 3 Report

There are some minor checks required. Unify the "thousands separators" - use either non-breaking (hard) space or comma. Line 95: In "the" first part instead of "a first part". Line 212: IRM acquisition "and" prior instead of "an prior". Lines 444, 452 and 550: Cercis siliquastrum should be in italics. Line 586: "they could have been taken" instead of: "they could have taken". Line 592: "long enough" instead of "enough long". 

I would consider adding a description of the legend (above the values) to the figure 5 itself.

I have marked the mistakes (in yellow) in the attached pdf.

 

 

Comments for author File: Comments.pdf

Author Response

There are some minor checks required. Unify the "thousands separators" - use either non-breaking (hard) space or comma. Line 95: In "the" first part instead of "a first part". Line 212: IRM acquisition "and" prior instead of "an prior". Lines 444, 452 and 550: Cercis siliquastrum should be in italics. Line 586: "they could have been taken" instead of: "they could have taken". Line 592: "long enough" instead of "enough long". 

Letaïef and co-authors: We thank the referee to point out some typographical errors persisted. We obviously corrected all the mistakes. We would also like to thank the rewiewer for his help to improve our manuscript.

 

I would consider adding a description of the legend (above the values) to the figure 5 itself

Letaïef and co-authors: We agree with this comment, the figure 5 has been improved as necessary.

Round 2

Reviewer 1 Report

In their cover letter, Letaïef and coauthor correctly state that my previous reviews did not provide a counter argument to improve linkages between the biomagnetic monitoring and CFD modeling described in the article.  Because my critique focused mainly on the linkages between the biomagnetic monitoring and CFD modeling rather than techniques for each, I'm willing to recommend that the article be accepted in its current form.  I do believe that the biomagnetic monitoring and CFD modeling described in the article may be of interest to researchers in those respective fields and potentially to the wider community of readers of Atmosphere.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This in depth-study of the effect of motorway barriers on the distribution of pollutants presents a very interesting and novel comparison between data obtained from chemical and magnetic measurement on the one hand, and a fluid dynamic computation of advective and turbulent pollutant dispersion under prevailing winds on the other. The combination of measurements and modeling approaches to such complicated system is highly valuable. I have only minor comments and suggestion for improving certain parts of the presentation of the CFD model, outlined below:

Line 44: You might add the effect of adsorption of pollutants on iron oxide nanoparticles, which exacerbates health issues.

Line 103: Daily average.

Line 189: Is the holder IRM stable over the time interval between holder pretreatment and IRM acquisition of the specimens? How large is the holder IRM with respect to the holder + specimen IRM? The method is only applicable if the holder IRM does not largely exceed that of the specimen. Furthermore, in my experience, holders can be very heterogeneous, with a fraction of them being excessively magnetic (e.g. by contaminations inside the molding). Have the holders being checked individually for such outliers before using them?

Line 206: It is better for the general understanding if individual terms in eq. (1) are explained first (i.e. from left to right: change in time, advective transport, diffusive transport, source term, sink term due to settling) followed by an explanation of the coefficients.

Section 2.4.2: This section is extremely difficult to understand without visualization, even to somebody with a good background in physics. I therefore suggest to move Fig. 7 to this point, because the calculation of the mean velocity field and turbulence is part of the method, rather than being a result. Adding the mean turbulent diffusivity profiles to this figure is another great help for better understanding how the model is constructed.

Line 268: “Therefore…thus” is redundant.

Line 386: “Note that if we anchored the regression line to the origin […] the magnetization must be equal to zero” does not make sense. IRM must be zero when MTE is zero, but this is exactly what anchoring to the origin means.

Line 388: The coefficient of determination increases when constraining the fit to the origin simply because there is only one free parameter instead of two, and does not indicate any model improvement. I would remove this consideration since it is misleading and does not change anything about the scatter of IRM data. Even if you discuss later the meaning of IRM measurements and the dispersion of IRM data, you should anticipate here one point that is missing in the whole manuscript, namely that IRM is a practical proxy of bulk magnetic contaminant concentration, forced by the instrumentation availability, but not the best one, which would be saturation magnetization. In practice, IRM is sensitive also to other parameters, notably domain state and stability. If you have any information on those aspect, this is the point to discuss it.

Line 412: Greater by at least…

Figure 5: The color scale is entirely dominated by two samples in the red range, so that other variations are subdued. Use a logarithmic scale instead.

Line 464: I would call these “parallelepiped rectangles” tubes with 5x10 m rectangular cross section running along each lane. Would be good if these are somehow visualized in Fig. 7.

Figure 8: I would label the vertical axes as “height above ground (m)”. The different curves represent a cross section somewhere at the location of the noise barrier; it would be helpful to specify this explicitly in the text and maybe even draw it in Fig. 1. Since there is some discussion about the role of barrier geometry, I would calculate one profile for one of the velocity for the case of earth-berms on both sides. The windward side should look different!

Line 550: I hope you can exclude holder problems. Some discussion about the fact that IRM is not an absolute measure of concentration is needed. Some studies report a huge IRM increase if measurements are performed at low temperature (see Sagnotti et al., 2009, Compositional, morphological, and hysteresis characterization of magnetic airborne particulate matter in Rome, Italy, G3 vol. 10, doi:10.1029/2009GC002563).

Line 586: Another point to consider is the fast weathering of pollutants: metallic powders from brakes are readily oxidized in soils.

Author Response

Line 44: You might add the effect of adsorption of pollutants on iron oxide nanoparticles, which exacerbates health issues.

Letaïef and co-authors:  We did not find in the literature conclusions about deleterious health effect specifically due to iron-oxides. Only health effects of metals are frequently reported. However, we changed the sentence with a more precise wording.    

Line 103: Daily average.

Letaïef and co-authors:  Done

Line 189: Is the holder IRM stable over the time interval between holder pretreatment and IRM acquisition of the specimens? How large is the holder IRM with respect to the holder + specimen IRM? The method is only applicable if the holder IRM does not largely exceed that of the specimen. Furthermore, in my experience, holders can be very heterogeneous, with a fraction of them being excessively magnetic (e.g. by contaminations inside the molding). Have the holders being checked individually for such outliers before using them?

Letaïef and co-authors:  Every care has been taken to ensure the accuracy of the measurements. The objective here is to subtract properly the sample holder signal. Because IRM can be imparted to the samples only when they are placed into the sample holders due to the sample nature (fresh leave, sifted soils, or paper filters), the protocol involves two steps. Two consecutive 1T-IRM were imparted and measured first to the empty holders and then to the sample holders once they were filled with the samples. The difference between these two measurements yield the sample magnetization assuming that the sample holder magnetization imparted in the first and in the second step remains constant. The sample holders are likely to carry a magnetization to a greater or lesser extent given the possible presence of impurity contents in the plastic molding or dust deposition on the sample-holder surface.

As stated by the referee, a key prerequisite in the protocol used is the reproducibility in the IRM acquisition by the sample-holder in the two steps. Two reasons could be behind a change in their magnetization. The first is a contamination by a deposition of dust on the sample-holder surface after the first measurement. This was easily controlled by means of a proper storage in a closed and clean container between the two measurements. In addition, we have made every effort to reduce the time elapsed between the two measurements, usually done in the same day, in order to reduce the chance of dust contamination on surfaces. The second is a strong viscous effect due to the presence of iron-oxide impurities in the plastic molding of a specific size to yield either a superparamagnetic or a multi-domain like behavior. Of course, we checked this physical property prior the experiments and found, as that is very often the case when an IRM is imparted on any material, a small viscous effect that lasts in the present case at maximum 2 minutes and 10 minutes for the empty sample-holder and the filled sample-holder, respectively. Hence, samples were placed in a zero-field environment during at least 15 minutes before to be measured. Such precautions are standard routines in all rock magnetism laboratories. The text has been completed in this sense. We had also a figure (Figure A1) which illustrates the magnetic viscous effect over the time for the sample holder and the filled holder sample.

Line 206: It is better for the general understanding if individual terms in eq. (1) are explained first (i.e. from left to right: change in time, advective transport, diffusive transport, source term, sink term due to settling) followed by an explanation of the coefficients.

Letaïef and co-authors:  We have changed the presentation of equation 1 and developed the text as recommended.

Section 2.4.2: This section is extremely difficult to understand without visualization, even to somebody with a good background in physics. I therefore suggest to move Fig. 7 to this point, because the calculation of the mean velocity field and turbulence is part of the method, rather than being a result. Adding the mean turbulent diffusivity profiles to this figure is another great help for better understanding how the model is constructed.

Letaïef and co-authors:  We are sorry that this part was enough clear. We have done our best to describe the model construction with sufficient details, clarity and lack of ambiguity to allow duplication of the results. Note that as already stated in the text, all the openFoam cases used in this work are available on-line as open access files at the following address:

https://zenodo.org/record/3961496#.X0aEdS1PiYU

This link allows everyone to reproduce the modeling. We do not think that moving Figure 7 to the model description may help in the understanding of the model construction because this figure illustrates a first important result in the model construction and thus as to be presented with results. However, we believed that a new figure illustrating the 4 rectangular cuboids used as volumetric sources for the traffic momentum and turbulence in the modeling of the stationary wind velocity field and in the passive scalar equation as volumetric sources for PM emission could help to a better understanding of the model construction. We add this figure as supplementary material (Figure A2).

Line 268: “Therefore…thus” is redundant.

Letaïef and co-authors:  Done.

Line 386: “Note that if we anchored the regression line to the origin […] the magnetization must be equal to zero” does not make sense. IRM must be zero when MTE is zero, but this is exactly what anchoring to the origin means.

Letaïef and co-authors:  We totally agree with this remark that pinpoints rather a problem of writing and more exactly a problem of awkward punctuation in the original version of the manuscript than a serious misinterpretation. However, following the next remark, we realize that it is wrong from a physical point of view to use the coefficient of determination r-squared to check the correlation between the two variables. Pearson’s and Spearman’s coefficients, already provided in the manuscript, are sufficient to enable us to highlight the good correlation between the metal concentration and the 1T isothermal magnetization. In the corrected version, we have dropped all the text related to r-squared and changed the Figure 3 removing the regression lines.

Line 388: The coefficient of determination increases when constraining the fit to the origin simply because there is only one free parameter instead of two, and does not indicate any model improvement. I would remove this consideration since it is misleading and does not change anything about the scatter of IRM data.

Letaïef and co-authors:  Done. See response above.

Even if you discuss later the meaning of IRM measurements and the dispersion of IRM data, you should anticipate here one point that is missing in the whole manuscript, namely that IRM is a practical proxy of bulk magnetic contaminant concentration, forced by the instrumentation availability, but not the best one, which would be saturation magnetization. In practice, IRM is sensitive also to other parameters, notably domain state and stability. If you have any information on those aspect, this is the point to discuss it.

Letaïef and co-authors:  We cannot but agree with this remark. Measuring a SIRM with a VSM/AGM under a strong field allows to eliminate instability in the magnetization due to the magnetic viscosity of multi-domain or superparamagnetic grains. However, sample preparation for VSM/AGM measurement is linked with numerous disadvantages compared to the protocol used in the present study.  Samples are of very small size (about 0.2 cm³) compared to the average size (about 100 cm²) of the samples we used, a surface corresponding to about 3-5 leaves of Cercis siliquastrum. We recall here that 5 samples by shred/tree are measured and then averaged. Doing thus each measurement reported on Figure 5B is obtained from 15-25 leaves. This also applies to the paper filters (area of 50.6 cm²). It appears much more difficult to reach such representativeness with the VSM/AGM measurements unless samples are prepared from leaching the leaves with distilled water and filtering the solution with the problem of recovering particles trapped in the epidermis of the leaves. In this case, the sample preparation would be time-consuming which is a limiting factor concerning the number of measurements carried out. In addition, numerous handling required in such sample preparation can alter the amount of dust deposition on leave surface. We have checked the time lapse of instabilities due to viscous effect and found that the magnetization remains enough stable after few minutes. The text was completed with these precisions. 

Line 412: Greater by at least…

Letaïef and co-authors:  Corrected.

Figure 5: The color scale is entirely dominated by two samples in the red range, so that other variations are subdued. Use a logarithmic scale instead.

Letaïef and co-authors:  The color scale is already not linear. We decided to keep it unchanged because we find that a log scale, at first sight, can appear to be more difficult to understand.

Line 464: I would call these “parallelepiped rectangles” tubes with 5x10 m rectangular cross section running along each lane. Would be good if these are somehow visualized in Fig. 7.

Letaïef and co-authors: We added as supplementary Figure A2, a new figure illustrating the 4 rectangular cuboids used as volumetric sources for the traffic momentum and turbulence as well as the volumetric source term emission for the traffic-related PM.

Figure 8: I would label the vertical axes as “height above ground (m)”. The different curves represent a cross section somewhere at the location of the noise barrier; it would be helpful to specify this explicitly in the text and maybe even draw it in Fig. 1. Since there is some discussion about the role of barrier geometry, I would calculate one profile for one of the velocity for the case of earth-berms on both sides. The windward side should look different!

Letaïef and co-authors: Ok for the y-axis labels. The studied profile is now indicated on Figure 1. Of course, we agree that it would have been interesting to also simulated the PM concentration for a profile south to the bridge in order to check the case with earth-berms on both sides. Unfortunately, we do not have a precise topography for this area unless to come back to the field to precisely measure the topography astride the carriage ways. We think that the main problem with our manuscript comes from the title which appears to us today, in regard to the referee’s comments, to be somewhat inappropriate. Indeed, our objective was more to check to what extend data from magnetic biomonitoring may be used as further information to validate CFD simulations on a real test case rather than to assess the berm geometry to mitigate the pollution related to the traffic. We have changed the title and improve the manuscript consistently.  

Line 550: I hope you can exclude holder problems. Some discussion about the fact that IRM is not an absolute measure of concentration is needed. Some studies report a huge IRM increase if measurements are performed at low temperature (see Sagnotti et al., 2009, Compositional, morphological, and hysteresis characterization of magnetic airborne particulate matter in Rome, Italy, G3 vol. 10, doi:10.1029/2009GC002563).

Letaïef and co-authors: We have clarified the text as suggested and added a supplementary figure (Figure A1) showing a small instability in the IRM signal due to a superparamagnetism behavior. Just please note that we were not able to reproduce the observation of Sagnotti et al. The signal of low-field magnetic susceptibility of our samples is too weak to be detected by our apparatus (Kappabridge KLY-3S) at low- and high-temperature.   

Line 586: Another point to consider is the fast weathering of pollutants: metallic powders from brakes are readily oxidized in soils.

Letaïef and co-authors: We are not sure we well understood your remark. Whatever the source of emissions (exhaust or wear products) the iron is readily oxidized to form in most cases magnetite which is systematically detected in magnetic biomonitoring survey of traffic related pollutants (see Hoffman et al., 2017 for example). For the other non-magnetic metals, XRF measurements do not allow a speciation and thus the degree of oxidation is not known.

Reviewer 2 Report

In this article, Letaïef and coauthors describe an measurement and modeling study of traffic-related particulate matter along a motorway with both earthen berms and noise barriers.  The measurement part of the study utilized multiple types of samples including from soil, plant leaves, artificial filters, and PM sensors to quantify the traffic-related particulates in front of and behind the different types of berms. The modeling part of the study utilized a CFD model to simulate the concentrations by height near the berms and noise barrier.  Despite these efforts, I don't think that the article answered the questions that many readers will have regarding these berms/barriers, including 1) which type of berm/barrier reduces particulates most effectively for near-road residents/businesses and 2) do berms/barriers lead to unsafe levels of particulates for drivers using the motorway, and 3) what are the dimensions of earthen berms or noise barriers that governments should build to reduce pollution near the motorway.  I'd recommend the authors revise the article to clearly answer these or other relevant questions and then structure the article to provide evidence for the answer.  I've also attached a pdf in which I've highlighted typos or grammatical errors that should be corrected in an updated manuscript.

Comments for author File: Comments.pdf

Author Response

In this article, Letaïef and coauthors describe and measurement and modeling study of traffic-related particulate matter along a motorway with both earthen berms and noise barriers.  The measurement part of the study utilized multiple types of samples including from soil, plant leaves, artificial filters, and PM sensors to quantify the traffic-related particulates in front of and behind the different types of berms. The modeling part of the study utilized a CFD model to simulate the concentrations by height near the berms and noise barrier.  Despite these efforts, I don't think that the article answered the questions that many readers will have regarding these berms/barriers, including 1) which type of berm/barrier reduces particulates most effectively for near-road residents/businesses and 2) do berms/barriers lead to unsafe levels of particulates for drivers using the motorway, and 3) what are the dimensions of earthen berms or noise barriers that governments should build to reduce pollution near the motorway.  I'd recommend the authors revise the article to clearly answer these or other relevant questions and then structure the article to provide evidence for the answer.  I've also attached a pdf in which I've highlighted typos or grammatical errors that should be corrected in an updated manuscript.

Letaïef and co-authors: The grammatical and typos mistakes highlighted by the reviewer have been corrected. About the bibliography the referee found inchoate, our introduction develops a state of the art on a very vast topic, which are the air quality and its challenges. We tried as best as could be cited the more 24 pertinent paper we found only in the introduction part. Of course, we aware that we may miss something. If so, we thank the reviewer to tell us which reference we did not cite and we will be pleased to read it.

We understand the reviewer’s questions about the role of these earth berms and wall barriers on the public health-care. Admittedly the objectives of this study were not explained clearly enough and will cause a misunderstanding about our intentions. Indeed, in this paper we want more to check if the magnetic biomonitoring combined with geochemical measurements can be used to validate and parameterize CFD simulations in a real study case rather than to evaluate the mitigation potential of the berm/barrier geometry and configuration on the traffic-related pollution. Regarding the different comments made by the reviewers we also realized that the title of this paper was confusing and somewhat inappropriate. In these perspectives, the title has been corrected and the manuscript has been improved.