Freshwater lake to salt-water sea causing widespread hydrate dissociation in the Black Sea

Gas hydrates, a solid established by water and gas molecules, are widespread along the continental margins of the world. Their dynamics have mainly been regarded through the lens of temperature-pressure conditions. A fluctuation in one of these parameters may cause destabilization of gas hydrate-bearing sediments below the seafloor with implications in ocean acidification and eventually in global warming. Here we show throughout an example of the Black Sea, the world’s most isolated sea, evidence that extensive gas hydrate dissociation may occur in the future due to recent salinity changes of the sea water. Recent and forthcoming salt diffusion within the sediment will destabilize gas hydrates by reducing the extension and thickness of their thermodynamic stability zone in a region covering at least 2800 square kilometers which focus seepages at the observed sites. We suspect this process to occur in other world regions (e.g., Caspian Sea, Sea of Marmara).

This paper discusses gas hydrate destabilization in the Black Sea. The authors posit that the known methane hydrate reservoirs in the seafloor sediment there are dissociating, and will continue to do so, as a result of increasing pore water salinity. They proceed to test this hypothesis by comparing the results of an equilibrium model with seismic (BSR) and other data.
I found the paper to be interesting and the results plausible. It is reasonably well-written and I had no problem with the grammar or logical progression. Having worked on the effects of ionic salts on hydrate stability in multi-component systems, the inhibiting influence of salinity is obvious and the unique, natural systems presented by the Black Sea with respect to a slowly-evolving solvent (i.e., liquid phase) field provides a good opportunity for in situ observation of salt-driven hydrate decomposition. It is refreshing to read a paper that views the stability of hydrate reservoirs through a more comprehensive thermodynamic lens that considers concentrations as well as merely temperature and pressure.
The field data appear to be of good quality, but seismics is not my area of expertise. Since the modeling results are a key component in testing the hypothesis, however, I feel that the manuscript would benefit from a few minor additions to address some omissions and make it selfcontained (i.e., complete). Specifically, in the Methods section on the topic of Numerical Model of GHSZ (lines 200 to 217), please include a few comments on how the endothermicity of decomposition and associated release of fresh water--which locally affects temperature and salinity, respectively, in this slow diffusive system--is handled. Perhaps these effects were found to be trivial relative to the salinity influx and thermal gradient over the timescales considered. If so, then please comment accordingly, since such self preservation phenomena are of interest to the readership.
In addition to the above, I suggest some small changes in wording to avoid the appearance of hyperbole. Notably, in Lines 122-123: "...the amount of methane released could reach one order of magnitude greater." This is rather quantitative. How was this estimated? If there are no calculations to justify the increase by an order of magnitude, then it should be re-worded. Also in Lines 126-128, the authors speculate about possible impacts related to tsunami generation and changes to the ocean carbon budget. Does the bathymetry suggest that a slump has a reasonable chance of generating a significant tsunami effect? Also, while the carbon budget in the Black Sea The salt diffusion model proposed is convincing. In this case, the authors should however describe what the limitations of the model are and explain the approximations applied. How can we approximate with one single value for water temperature and a geothermal gradient based on the first 12m of sediments below the seabed the evolution of a basin in a time interval of 5k years? And the effects of erosion? Sedimentation rate? It is important to mention that most of the data use here is potentially linked to the presence of submarine canyons.
The extrapolation of the observation from the study area to the entire Black Sea is not convincing. The authors not only assume a nearly homogenous distribution of GH but also that the Temperature-Pressure-Salinity conditions are homogenous. This make the first order constrain of the GH destabilization zone very weak.
I do encourage the authors to check carefully the literature about the other basins before mention them as potential analog of the Black Sea.
I suggest the authors to provide the appropriate information in order to validate the method or to consider rewriting the manuscript in an area where the key information are available.
The manuscript is well-written and reads clearly. There are no ethical concerns present. The figures are clear and well edited.

Detailed comments
Lines 9-18 (Abstract): I suggest revisiting the abstract.
Lines 19-31: the paragraph is well-written but contains too much information on why gas hydrates (GH) are important and less on the general aspects of GH. I suggest rewriting this paragraph and trying to get the reader ready to get the rest of the manuscript.
Lines 32-35: the BSR, as discussed by Xu and Ruppel 1999, does not coincide to neither the base pf the gas hydrate stability zone (GHSZ) nor the top of the free gas zone (FGZ). I suggest referring to this paper for a better significance of the BSR. I would explain also what a negative polarity is. It is negative with respect to what. If the polarity of the seismic data is European the BSR is positive.
Lines 36-38: provide a reliable distribution map of the BSR, which you will use then to infer the distribution of the GH. Without the distribution of the GH in the Black Sea, most of your argument from line 112 onwards are hard to argue (even for a first order evaluation Lines 58-59: please specify if this gas is coming from dissociation of GH or free-phase gas. Lines 63-63: the salinity is measure to a depth of 25m How you know that in the deeper section (e.g. at the BSR) higher salinities are not present? Reference 29 ( fig. 5 top) shows indeed an increase of salinity with depth below 40m (The simulated dissolved Cl− profile retains the memory of marine-like waters of the last interglacial…). Provide a salinity profile for the entire GHSZ.
Lines 72-74: here you are using 1D geothermal gradient (based only on the first 12m of sediments!!!), one seawater temperature (!) and 99% methane to run simulations which you then apply to the entire Black Sea! Please explain all the limitations.
Lines 78-79: divergence of BSR predicted/observed are probably the result of a constant geothermal gradient based on the first 12 m of sediments. Revisit this interpretation once the GHSZ is calculated more precisely.
Lines 90-93: please be specific on the limitation of the method. Sedimentation rate? Effect of freshening of rivers? Erosion? Lithology? Your study area and most of the derived data is nearby a submarine canyon. What does this imply?
Lines 94-111: to discuss again once all the points suggested above are taken into account.
Lines 122-130: I suggests revisiting this once you have provided enough information to calibrate your model. As it is now this is not adequate even for a first order constrain.
Lines 131-133: provide adequate references, as some of the cited ones (Reference 28 and 41) do not mention the effect of salinity on the destabilization of GH.

Subject: paper NCOMMS-16-26092 -Response to reviewers.
Reviewer 1: 1) Abstract should be revised because it is hard to understand what it is exactly done in this study with current abstract.
We have partly rewritten and reorganized the abstract taking into account this comment. See lines 11 to 18 of the revised manuscript.
The table below shows the composition of the gas sampled in the study area. The 5% are justified by the amount of hydrate recovered from the three cores sampled in the study area which was about 10% of the whole cores. However, and based on previous published data, we introduced in the revised manuscript a hydrate concentration between 1% and 5% of the porosity (60%). Details are added in lines 128 to 131 of the revised manuscript.
Reviewer 2: 1) In the Methods section on the topic of Numerical Model of GHSZ (lines 200 to 217), please include a few comments on how the endothermicity of decomposition and associated release of fresh water--which locally affects temperature and salinity, respectively, in this slow diffusive system--is handled. Perhaps these effects were found to be trivial relative to the salinity influx and thermal gradient over the timescales considered. If so, then please comment accordingly, since such self-preservation phenomena are of interest to the readership.
The numerical model used fully accounts for the latent heat effects which may impede gas hydrate dissociation (self-preservation phenomenon). Indeed, additional heat source and heat sink are produced as gas hydrate forms and dissociates respectively (Sultan et al., 2004). Gas hydrate dissociation is also known to cause a local decrease in salinity which may also impede the decomposition process. However, for the considered timescale calculations (kyrs), the small salt perturbation occurring at the hydrate border is expected to be second order with respect to the general process generated by the vertical salt diffusion. This part was added to the manuscript lines 217 to 223 of the revised manuscript.

It is true, we don't have a numerical estimation to justify our purpose although we consider that the hydrate decomposition and the subsequent slope instabilities may change the T/P/S limit conditions and therefore amplify the hydrate decomposition process (Maslin et al., 2010). We agree with the reviewer that this sentence is more speculation than demonstration and if necessary can be removed from the text.
3) Also in Lines 126-128, the authors speculate about possible impacts related to tsunami generation and changes to the ocean carbon budget. Does the bathymetry suggest that a slump has a reasonable chance of generating a significant tsunami effect? 4) Also, while the carbon budget in the Black Sea appears to be currently affected by methane release from hydrates, this is a local phenomenon that may or may not change significantly over time. I doubt that salt-driven hydrate destabilization will impact the global carbon budget. Reviewer 3:

Studies on recent and historical tsunamis recorded in the Black
1) The claims are not however convincing because of the lack of data to constrain the salinity at the BSR. The salinity profile is indeed limited to the first meters of sediments and far away from the BSR. This preclude to understand whether the GH are in a stable conditions. The authors should provide evidence of this. If the salinity at the BSR is close to the one of the seabed the GH are stable and will not dissociate in the near future. 2) The salt diffusion model proposed is convincing. In this case, the authors should however describe what the limitations of the model are and explain the approximations applied. How can we approximate with one single value for water temperature and a geothermal gradient based on the first 12m of sediments below the seabed the evolution of a basin in a time interval of 5k years? And the effects of erosion? Sedimentation rate? It is important to mention that most of the data use here is potentially linked to the presence of submarine canyons. 3) The extrapolation of the observation from the study area to the entire Black Sea is not convincing. The authors not only assume a nearly homogenous distribution of GH but also that the Temperature-Pressure-Salinity conditions are homogenous. This make the first order constrain of the GH destabilization zone very weak. This has been done and the three others basins mentioned in the study may undergo the same dissociation process due to salt diffusion. We have added some references showing the presence of hydrates at the landward termination of the GHSZ and the salinity gradient of these three other areas.

The calculation of the mean geothermal gradient using the depth of the current BSR in the Black
It may be noted that we have enriched the manuscript with 17 references.

Reviewers' comments:
Reviewer #2 (Remarks to the Author): After reviewing the revised manuscript and the authors' rebuttal to my earlier comments, I am satisfied with their responses and believe that the manuscript warrants publication Reviewer #3 (Remarks to the Author): The paper claims that diffusion of salt in pore space in the first meters of sediments triggered by changes in salinity of the seawater can destabilise the gas hydrates with a subsequent release of methane into the atmosphere. The idea is novel and of great interest however the extrapolation of the idea to the entire black sea is still not convincing. I suggests to review the amount of gas release from the gas hydrate destabilisation zone once a clear distribution of gas hydrates and salinity is taken into account. I prepared a document (attached) with comments not addressed in the previous revision and new comments based on this version of the manuscript.
Reviewer 3: Response to comments not addressed from the previous review Lines 32-35: the BSR, as discussed by Xu and Ruppel 1999, does not coincide to neither the base pf the gas hydrate stability zone (GHSZ) nor the top of the free gas zone (FGZ). I suggest referring to this paper for a better significance of the BSR. I would explain also what a negative polarity is. It is negative with respect to what. If the polarity of the seismic data is European the BSR is positive. Line 39: Reference 28 shows evidence of freshening consistent with a dehydration caused by smectite to illite process.

Indeed the authors of this paper show that the source of fluid freshening in the central part of the Ulleung Basin (Japan Sea or East Sea) is due to dehydration caused by smectite to illite process. This process was not described in the Black Sea.
Lines 47-49: a map showing more information of the actual salinity of the Black Sea is needed. What is the impact of the rivers on the salinity of the Black Sea? You cannot extrapolate your model on the entire Black Sea.

Water exchange is low between the Black Sea and the Mediterranean Sea due to the narrow Bosphorus
Strait. Fresh water entering the Black Sea by river discharge is less dense than water from the Black Sea, mixing of fresh water/sea water takes place in the upper part of the water column, above 90 meters. Under the pycnocline (the water density limit that restricts vertical mixing and exchange between the deep layers and the mixed layers), the water column of the Black Sea is completely anoxic. The salinity measurements in the water column of the Black Sea is close to 22 psu above 300 m water depth (e.g.

Reviewer 3: Response to comments on the revised manuscript
Line 45-49: as already commented on the previous version of this manuscript a detailed map of the seabed salinity is necessary before to calculate the amount of CH4 released from the destabilization of the gas hydrates. Although the authors provide evidence of low salinity at site DSDP 379, at DSDP 380 and 381 the salinity is very different (e.g. from Manheim, 1978): These values show that the salinity is not so low below the seabed and that the gas hydrate are stable! Also, it would be interesting to evaluate whether the proposed GH destabilization process is more impactful than other documented gas expulsion phenomena deriving from other processes. For the considered 5 kyrs calculation period, we considered the impact of an ongoing proven process related to salt diffusion and we considered unchanged bathymetry and constant temperature and pressure conditions (Past Interglacials Working Group of PAGEs, 2016). It was possible to consider in the calculation a seabed temperature or a sea-level change but how to evaluate this T-P evolution over a 5 kyrs calculation period?
Reviewers' comments: Reviewer #3 (Remarks to the Author): The paper claims that diffusion of salt in pore space in the first meters of sediments and triggered by changes in salinity of the seawater can destabilise gas hydrates with a subsequent release of methane into the atmosphere. The idea is novel and of great interest however, as already explain in the previous occasions, the application of the model to the entire black sea is still not convincing. I suggests to review the amount of gas release from the gas hydrate destabilisation zone once a clear distribution of gas hydrates is taken into account. I prepared a document (attached) with few detailed comments based on the latest version of the manuscript. The polarity of the BSR is negative only if the polarity of the dataset is American (where a decrease of acoustic impedance with depth is represented by a negative reflection coefficient); if the polarity of the dataset is European (where, instead, a decrease of acoustic impedance with depth is represented by a positive reflection coefficient) the BSR is positive. When you say "The BSR represents the base of the Gas Hydrate Stability Zone (GHSZ) that appears as strong, negativepolarity,..." is a valid statement, but you have to define the polarity of the dataset first.

Answerpresent version
This is correct, we use the American convention (a decrease of impedance is represented by a negative reflection coefficient). This is now mentioned in the caption of figure 2.
Previous reviewer comment/answer/reviewer comment (2) Lines 36-38: provide a reliable distribution map of the BSR, which you will use then to infer the distribution of the GH. Without the distribution of the GH in the Black Sea, most of your argument from line 112 onwards are hard to argue (even for a first order evaluation). Reference 1 (fig. 4) (2016), the map presented in Fig. 6 shows the volume of CH4 in hydrates in standard conditions for the Black Sea suggesting the occurrence of hydrates in our region of interest where hydrate may dissociate by salt diffusion. The Reference 26 (Popescu et al., 2006) shows areas covered by BSR in NW Black Sea where 2D seismic data were available. The extended data acquired in the GHASS project in 2015, some of which are shown in our study, show that the BSR is observed in all seismic data located to the north of the Danube canyon at a water depth in agreement with the base of the gas hydrate stability zone. The cartography of this BSR is not the aim of the present paper but it may be added to the supplementary data if you consider it as essential. This is the distribution of BSR on the Black Sea from Vassilev, A., (2006). This distribution map is based on a work of Popescu et al., 2006, which include only 5 areas and 11 seismic segments (thick black lines). The other maps showed in the same work are based on estimated optimistic and pessimistic scenarios and not on observed BSR. Merey and Sinayuc, 2016 used estimated values. It is also worth noting that previously published maps based on factual observations of BSR in the Black Sea have not taken into account, e.g.: Poort et al., 2005;Vassilev & Dimitrov, 2003;Merey et al, 2016. Again, the authors are invited to reconsider the risk of extending the application of their model for the destabilization of the GHSZ to the entire Black Sea.

Answer present version
In the previous version of the paper we calculated the upper bound of the area where the gas hydrate may decompose due to salinization process. In this upper-bound calculation we assumed that gas hydrate is occurring in the whole Black Sea wherever the thermodynamic conditions are valid. We agree with the reviewer 3 about the possibility to improve the previous extrapolation. In the present version of the paper, the new calculation considers only areas where free gas/gas hydrates have been recovered or inferred (see Supplementary Fig. 2 and references in caption). We have thus modified our estimation of the amount of gas generated by gas hydrate dissociation due to salinization of sediment. Figure 1

Previous
reviewer comment/answer/reviewer comment (3) Line 39: Reference 28 shows evidence of freshening consistent with a dehydration caused by smectite to illite process. Indeed the authors of this paper show that the source of fluid freshening in the central part of the Ulleung Basin (Japan Sea or East Sea) is due to dehydration caused by smectite to illite process. This process was not described in the Black Sea.
The dehydration from smectite to illite produce fresh water and not chlorine rich water. Fresh water do not destabilizes GH. The Ulleung Basin cannot be used as analogue.

Answer present version
Yes we agree, the Ulleung basin is not a good example. We have now removed all the references to this area from the manuscript.

Previous
reviewer comment/answer/reviewer comment (4) Also, it would be interesting to evaluate whether the proposed GH destabilization process is more impactful than other documented gas expulsion phenomena deriving from other processes. For the considered 5 kyrs calculation period, we considered the impact of an ongoing proven process related to salt diffusion and we considered unchanged bathymetry and constant temperature and pressure conditions (Past Interglacials Working Group of PAGEs, 2016). It was possible to consider in the calculation a seabed temperature or a sealevel change but how to evaluate this T-P evolution over a 5 kyrs calculation period? Apologies if I was not clear. What I was trying to ask was related to the effective amount of methane released by the salt diffusion model with respect to other processes of methane release in the Black Sea. For instance, the author show release of methane from a region where the GHSZ is expected to be stable (e.g. Figure 4) and suggest that some amount of gas could have been released from faults (Supplementary Figure 7). So, what is the impact of such processes with respect to the salt diffusion model proposed?

Answer present version
Our work has allowed to determine a rough indication of the methane released by salt diffusion. Without an important plan for monitoring free gas fluxes at the seafloor level inside but also and especially outside the GHSZ it will be very difficult to evaluate the ratio between methane released by the salt diffusion with respect to other processes. This question could be the starting point of an ambitious future research project in the Black Sea. On the other hand, the impact of the local fault system as a preferential path for gas migration is expected to be negligible because it is an isolated structure and almost unique in our study zone (Riboulot et al.,  2017).