the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Dec 2023
Probabilistic hazard assessment of the gas emission of Mefite d’Ansanto, Southern Italy
Abstract. The emission of gas species dangerous for human health and life is a widespread source of hazard in various natural contexts. These mainly include volcanic areas but also non-volcanic geological contexts. A notable example of the latter occurrence is the Mefite d’Ansanto area in the Southern Apennines in Italy. Here, large emissions of carbon dioxide (CO2) occur at rates that make this the largest non-volcanic CO2 gas emission in Italy and probably of the Earth. Given the topography of the area, in certain meteorological conditions a cold gas stream forms in the valleys surrounding the emission zone, which proved to be potentially lethal for humans and animals in the past. In this study we present a gas hazard quantification study that considers the main specie, that is CO2, and the potential effect of the most dangerous, which is hydrogen sulphide (H2S). For these purposes we used VIGIL, a tool that manages the workflow of gas dispersion simulations specifically optimised for probabilistic hazard applications. Results are discussed and presented in form of maps of CO2 and H2S concentration and persistence at various exceedance probabilities considering the gas emission rates and their possible range of variation defined in previous studies.
- Preprint
(1655 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2023-2867', Anonymous Referee #1, 14 Mar 2024
The article “Probabilistic hazard assessment of the gas emission of Mefite d’Ansanto, Southern Italy” presents several gas dispersion simulations applied to previous recorded data and constitute a valuable approach to evaluate the hazard in this area that releases anomalous high CO2.
Diffuse degassing areas may occur both in volcanic and non-volcanic environments, as Mefite d’Ansanto and these approaches may be applied in several other degassing sites. The article well describes the tools and the methodologies used to perform the simulations, what is a positive aspect for further users. The article is in general well organized and the figure/tables are adequate.
I would suggest, however, a better clarification of the newness of the article, since it is not so clear. Is there any step of the analyses that is new in the current study or it is just the first time is applied in this degassing area? All these procedures integrating different tools (DISGAS, TWODEE-2, DIAGNO) were already applied before? This should be better clarified. Authors estimate probability maps not only for the CO2, but decided also to include the H2S. Was this approach already applied in other sites? Considering that CH4 may also be hazardous in certain concentrations and that the CH4 concentration in the emission is also high (Table 1 from the article), did authors consider to apply the methodology also to the CH4?
Chiodini et al. (2010) applied the TWODEE-2 code to the same dataset at Mefite. Authors should better discuss what are the major differences obtained between the two different studies especially when DISGAS is applied to similar wind conditions as the ones applied in the previous study.
Some more general comments:
Section 1. I suggest authors to review the references used for the CO2 thresholds and exposure time. Authors refer in one case to an Italian reference (Settimo et al., 2022) together with some studies that do not focus on these thresholds but use them based on other literature. I suggest one of the options: or authors include the references and add “and references therein”, or, in a better way, use more fundamental and specific literature on these thresholds, impacts and exposure time. Some recommendations are: Blong, 1984; Wong, 1996; IVHHN, NIOSH - https://stacks.cdc.gov/view/cdc/19367 (line 21 – page 1). I also suggest checking the references sequence in the text. Some of them do not appear with the chronological sequence.
Section 2. On the characterization of the area I suggest authors to better characterize the area including mention that the emissions are cold and the fluxes previously estimated by Rogie et al. (2000) and Chiodini et al. (2010). For instance, Rogie et al. (2000) report that CO2 concentrations > 30 vol.% were measured at an height higher than 2 m. This is an interesting aspect to recall in the discussion to compare with the results obtained in the current study.
Section 4. Tables 2 and 3 in my opinion need to be improved. Another column should be added in both tables to mention the maximum recommended exposure time for each of the thresholds. I would split the effects and the exposure times. Then, the last column would report the “tested exposure time in the current study”. Otherwise it seems that humans can be exposed to 10 vol.% during 1 hour (Table 1) and this is not true and could even be lethal. In fact, in Table 2 authors need to add that in this last threshold (100 000 = 10 vol.%) death can also be one of the effects. Same comment for the 500 ppm associated to the H2S. Still in what concerns Table 2, I suggest authors to use specific literature (NIOSH, Blong, Wong, as mentioned above, and complete the symptoms per threshold considered, since there is lack of symptoms in the table). The IDLH mentioned in the text (line 259, page 10) for the H2S also exists for the CO2 and should be mentioned (and added in the table). I suggest to review this section based on additional literature.
Section 5. The discussion and evaluation of the hazard considering different seasons is a very interesting contribution that should be applied in other degassing areas. Nevertheless, it is important that future studies attempt to couple seasonal degassing maps with the meteorological data associated with the different seasons. Several studies on degassing areas showed that CO2 is usually higher during winter comparing to summer, and for this reason the evaluation of this coupled (and eventually contradictory) effect will be very interesting. Is this the first study that reports these seasonal maps? If yes, this is not mentioned in the abstract and it should.
The are some general technical comments that I would like to call the attention, namely the need to control all the figures and tables that do not appear correctly in the text. This needs to be carefully checked:
- For instance, in line 29 – page 1, authors refer Figure 1, but this sentence refers to Figure 2b.
- Line 78, page 3 – the links to the webpages of NIOSH could be added as references (e.g. NIOSH, 2019), and then the links appear in the references list.
- Line 85, page 3 – authors should add the units also for the isotopes in the table.
- Figure 1 was redrawn from Chiodini et al. (2010). Looking at the literature, there are several geophysical studies (e.g. https://www.mdpi.com/1424-8220/23/3/1630; https://pubs.geoscienceworld.org/ssa/bssa/article-abstract/113/3/1102/620681/Hydrothermal-Seismic-Tremor-in-a-Wide-Frequency?redirectedFrom=fulltext) that were recently developed and I wonder if they could be used to improve the scheme showed. Authors should check. Anyway, remember that Figures 1 and 2 need to be checked as they are wrongly referenced in the text.
- Line 97, page 4 – check the chronological sequence of the references.
- Line 204, page 8 – when appears “figure 2” should appear “Figure 2”.
- Figure 2 – I think Figure 2 could be improved by inserting in Figure 2b a square with the location of Figure 2c.
- Line 221, page 9 – check what figure should be mentioned in the text. It does not refer to Figure 1.
- Line 245, page 9 – the number of the tables also need to be checked along the text. I believe that authors refer to “Table 2” and not “Table 1” as it is written.
- Table 3 – there is a reference missing in the first line.
- Line 285, page 13 – I think authors meant “winds blowing” instead of “finds blowing”. Check.
- Lines 294 and 295, page 14 – I could not see in the figure the statement “is also not negligible… along the main”. Maybe I misunderstood the sentence, but please check.
- Line 307, page 14 – Table 4 instead of Table 3.
- Line 318, page 14 –I wonder why authors decided to check the exceedance probability of 16%, and not any other percentage? What was the criteria?
- Line 382, page 21 – authors mention that certain concentrations of CO2 may be very dangerous for the human health, but I would even add for the “human life”, as in certain concentrations CO2 is lethal and it was even reported casualties in the area.
- Line 404, page 22 – “was evaluated” is repeated in the sentence.
Citation: https://doi.org/10.5194/egusphere-2023-2867-RC1 -
AC1: 'Reply on RC1', Fabio Dioguardi, 22 Mar 2024
Dear editor and community,
we attach a pdf with our replies to the reviewer.
Fabio Dioguardi
-
RC2: 'Reply on AC1', Anonymous Referee #1, 25 Mar 2024
After reading the point to point answers, where authors state that they will implement the suggested corrections, I have not further comments. I am happy that the article is published after implementing the corrections.
Citation: https://doi.org/10.5194/egusphere-2023-2867-RC2 -
AC3: 'Reply on RC2', Fabio Dioguardi, 05 Jul 2024
We thank again the reviewer for the contribution.
Regards
Fabio
Citation: https://doi.org/10.5194/egusphere-2023-2867-AC3
-
AC3: 'Reply on RC2', Fabio Dioguardi, 05 Jul 2024
-
RC2: 'Reply on AC1', Anonymous Referee #1, 25 Mar 2024
-
RC3: 'Comment on egusphere-2023-2867', Matteo Cerminara, 04 Jun 2024
The article “Probabilistic Hazard Assessment of the Gas Emission of Mefite d’Ansanto, Southern Italy” integrates data from Rogie et al. (2000) and Chiodini et al. (2010) with the VIGIL python tool (Dioguardi et al. 2022) to evaluate the hazards posed by CO2 and H2S emissions in the Mefite d'Ansanto area. The manuscript is well-written and clearly explains the tool's functionality and its approach to managing and combining results from multiple simulated scenarios. However, improvements are needed in the descriptions of the physical models used, parameter choices, and the evaluation of epistemic uncertainties.
General Comments
The core models utilised by VIGIL are DISGAS and TWODEE. DISGAS is an advection-diffusion model that transports a passive scalar emitted from a steady source within a prescribed wind field, refined using the DIAGNO model. It disregards gravitational effects. TWODEE, on the other hand, is a shallow water model for transporting a gravity current that interacts with prescribed wind and topography. DISGAS is appropriate for diluted flows and strong wind conditions (Richardson number, Ri, much less than 1), whereas TWODEE is used when gravitational effects dominate (Ri much greater than 1). Intermediate regimes, which can exhibit behaviours divergent from these two models, are more complex. While the authors correctly introduce and discuss this point, they do not quantify the epistemic uncertainty introduced by these approximations. Comparing the two models in intermediate regimes can help quantify this uncertainty. The authors use DISGAS when Ri < 0.25 and TWODEE when Ri > 0.25. I recommend comparing CO2 gas concentrations obtained with TWODEE and DISGAS for a scenario with Ri ~ 0.25 and a wind field from W-SW to quantify the differences between the two approaches under challenging conditions.
This comparison is crucial also because the persistence values along the valley shown in Figure 4 likely result mainly from TWODEE simulations. The Richardson number, used to decide which model to apply, is calculated at the source. However, dilution due to entrainment and turbulent diffusion tends to decrease the local Richardson number, making the current lighter. Additionally, the Richardson number is averaged over a whole day, meaning some hours with Ri < 0.25 are modelled with TWODEE, which stretches TWODEE's capabilities in diluted regimes. I recommend adding a comment in the main text to describe these approximations.
Another critical aspect needing better description is the effect of turbulent diffusion. In advection-diffusion models like DISGAS, turbulent diffusion is a key parameter determining how much the flow dilutes and how gas concentration decreases with distance. This parameter is also important in TWODEE, alongside entrainment. The authors should explicitly explain how these parameters are calculated in the two models and how turbulent diffusion depends on mesh resolution. I suggest adding a map (even in the supplementary material) showing the depth-averaged horizontal turbulent diffusion used by both models for scenarios with median, strong, and weak wind conditions, using the paper’s resolution (3 m) and a refined resolution (1.5 m).
Simulations with refined resolutions are also essential for quantifying the epistemic uncertainty due to numerical approximations in this specific application. Including the effect of resolution in a supplementary figure would be beneficial.
All parameters used by the two models should be explicitly listed in a table to help readers understand the modelled conditions without needing to download the database and delve into the model configuration.
Meteorological conditions used in the simulations should be presented more clearly. The spatial resolution is 30 km, correct? Where is the center of the cell used for this specific application located? What is the temporal resolution (I assume it is 1 hour)? I recommend including this information in Section 3.2.1. Additionally, I suggest adding a figure (even in the supplementary material) showing an example of the vertical profiles downloaded from the cited datasets up to the height of the numerical domain. Highlight the value used for the quantification of the Richardson number and the Monin-Obukhov length scale, as well as the value used to draw the wind speed distribution shown in Figure 7 (at 10 m above the ground). How much variability does DIAGNO introduce to the original wind profile? I recommend showing the median and standard deviation of the vertical profiles obtained by applying DIAGNO to the example meteorological condition in the same figure.
It is unclear to me how many simulations are performed, how the source distribution is sampled, and how much computational time is required. Specifically, it appears that 24 simulations for each of the 1000 sample days are performed to account for the variability of meteorological conditions, resulting in a total of 24,000 1-hour simulations with fixed source conditions. How is the source variability taken into account? If the normal distribution introduced in the manuscript is sampled with, say, 10 points, the total number of simulations would be 240,000, correct? Please clarify these points in the main text.
Specific commentsWeb links: Numerous web links are present in the main text. Is it possible to move them to a specific reference section with the date of last access?
Zenodo Dataset: I was unable to access the Zenodo dataset.
Section 2: The source data are from several years ago (Chiodini et al. 2010 and Rogie et al. 2000). Please add a comment on the expected variability at the source after 14 years (for the mass flow rate) and 24 years (for the composition).
Line 113: Is the momentum coupling one-way or two-way? Please specify explicitly.
Section 3: Please include a comment about the potential chemical reactions affecting H2S during its transport and their time scales.
Lines 147-151: The difference between Forecast and Reanalysis mode is not perfectly clear to me. Why can't the ERA5 dataset be used in Forecast mode and NCEP in Reanalysis mode? Please clarify the differences between these two datasets.
Line 176: ECDF is not defined when first introduced. It is defined later at line 186.
Section 4.1: The height of the numerical domain and the vertical size of the cells are not reported.
Figure 3: Please add a box highlighting the emission area.
Figures 3, 4, 5, 6, 8: Please indicate, for each figure, how many scenarios come from TWODEE and how many from DISGAS simulations. This information is essential to understand the regimes producing hazardous conditions.
Line 285: Change "finds blowing" to "winds blowing."
Figure 5: It is not clear whether the curves represent 24-hour averages or hourly results.
Figure 4 and Comments in the Text: The method for calculating the persistence maps is not completely clear to me. For example, when you say CO2 persistence > 5000 ppm for 8 hours, does this mean selecting (cell by cell) all the 24-hour scenarios with 8 consecutive hours satisfying the condition? If the condition is satisfied for 4 hours, then concentration decreases for 1 hour and then increases again for 4 hours, do you keep or discard the scenario?
Figure 7: Link the figure to the discussion at line 225.
Line 387: It is unclear how a lower estimate in a probabilistic hazard map (Figure 4def) can be described as "safe."
Lines 402-403: It is not specified how many points were used to sample the source variability.
Lines 404-406: State explicitly that this approach disregards intra-day variability of the Richardson number.
Citation: https://doi.org/10.5194/egusphere-2023-2867-RC3 - AC2: 'Reply on RC3', Fabio Dioguardi, 03 Jul 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 394 | 97 | 44 | 535 | 27 | 25 |
- HTML: 394
- PDF: 97
- XML: 44
- Total: 535
- BibTeX: 27
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
