The Arctic Permafrost Vulnerability Index

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article is relevant and useful in theoretical and practical aspects.

lines 38-40. It should be clarified that only the upper horizons of permafrost are thawing, and in several regions the permafrost will disappear completely. It is necessary to distinguish between territories with the complete disappearance of permafrost and the permafrost disappearance from the domain of the designing and construction. The authors mentioned this in the "Infrastructure" section, but in other sections the concept remains ill-defined.

The "permafrost thaw" index does not meet the "scalability" requirement. The permafrost thawing was based on modeling the permafrost temperature in the selected landscape type. However, the diversity of heat exchange conditions through the surface leads to a diversity of permafrost response types to climate change [Shur & Jorgenson, 2007]. Thus, the regional permafrost thawing estimate does not yet have reliable information sources, unlike local forecasts, which are provided with a good variety of input data for modeling.

The index "exposure" actually corresponds to the "danger", because not every event caused by permafrost thawing leads to the damage. For example, the development of thermokarst in the tundra is a danger, but only if thermokarst develops under a road or a building, then we can talk about the occurrence of exposure. The authors allow logical liberties, interpreting permafrost thawing as the exposure. The permafrost will not thaw everywhere and simultaneously, but only in certain landscapes. Not everywhere where permafrost is thawing the unfavorable processes become active. Not everywhere where dangerous processes develop, their magnitude can lead to damage to human infrastructure.

"Exposure" and "Adaptive capacities" are functionally related when calculating the range of vulnerabilities (Figure 1). However, there are situations where "Low exposure" is adjacent to "Low capacity" or "Very high exposure" is adjacent to "Very high capacity".

When considering indices, authors are encouraged to discuss and analyze the following circumstances:

"Permafrost coverage" is not related to the danger to infrastructure. For example, continuous permafrost on a low ice content rocks is much less dangerous for construction than sporadic, rare, ice-bearing islands of permafrost on peatlands in valleys. Construction of a building on such peat island will lead to significant costs for engineering protection or a building repair. Even in areas with continuous permafrost the engineers choose the relatively favorable local areas. An analysis of spatial patterns of damage from permafrost processes showed its lack of dependence on the permafrost extent [Sergeev, Chesnokova, Morozova, 2015].

"Change in permafrost extent" is not provided with reliable data in practice, since the modeling is related only to a highly simplified set of initial data "Permafrost_cci" and is poorly coordinated with "Ground ice coverage" from the Brown's Circumpolar Map. Soil warming trends are informative only if there are data on ice content, moisture regime of the active layer, evaporation and knowledge of the temperature distribution in the lower horizons of permafrost. If these data are not available, then simple averaging of warming trends over the territory will give incorrect results.

Line 174. What the authors use should rather be designated as a "forecast of thermal subsidence", but not thermokarst. The development of thermokarst assumes a consistent and staged development of characteristic forms. Thermokarst coverage does not always correlate with the danger of thermal subsidence, since subsidence can occur, but thermokarst does not develop.

Line 162-165. Questionable conclusion. The highs ice-content leads to the slowing the permafrost thaws, but the final damage becomes greater. Such an index should only be used with a time scale in mind. The vulnerability index should be tied to the forecast period.

Line 225 The index appears to be incomplete and formal, since it lacks an indicator of differences in political management of adaptation in different countries. If in the Western Arctic the factor of self-government at the municipal level is significant, then in Russia adaptation is appointed by decree from Moscow and is carried out mainly with federal funds. This leads to different weights of the same indicators in different sectors of the Arctic. For example, half of the population of the gas field may have higher education, but there is no adaptation, because the boss did not order and did not allocate money.
The content and reasons for population migration in different places are also very different.
The economy in different countries is also structured differently. The income of the administrative territory may be high, but the budget structure may not allow for the development and implementation of adaptation plans.

Line 477 Authors are advised to specify what the vulnerability index can be used for. It is said that it is basic, but basic for what? What classes of management decisions need it? The generalized vulnerability index seems to be of little use due to the mixing of assessments that are heterogeneous in time and at territories. It seems that a more productive path is to select management decisions for paired combinations of particular indices.

 

Author Response

We appreciate your feedback and have addressed your comments below. Additionally, we have revised the manuscript to enhance the clarity of our work.

Comment 1: lines 38-40. It should be clarified that only the upper horizons of permafrost are thawing, and in several regions the permafrost will disappear completely. It is necessary to distinguish between territories with the complete disappearance of permafrost and the permafrost disappearance from the domain of the designing and construction. The authors mentioned this in the "Infrastructure" section, but in other sections the concept remains ill-defined.

Answer 1: Thank you for this comment. The difference between areas where permafrost will disappear completely and those where it will thaw only partially is considered in the exposure index, through the different permafrost thaw indicators. For example, the permafrost extent shows the annual probability of permafrost to occur. The data is available for each year between 1997 and 2019. We measure the differences between the extent in 2019 and 1997 to estimate the probability of permafrost thaw in each area (1x1 km²). Also we disagree with the fact that "only the upper horizons of permafrost are thawing". In many cases, abrupt thaw processes cause deep permafrost to thaw, which we wanted to take into account with the indicator reflecting the probability for thermokarst.

Comment 2:The "permafrost thaw" index does not meet the "scalability" requirement. The permafrost thawing was based on modeling the permafrost temperature in the selected landscape type. However, the diversity of heat exchange conditions through the surface leads to a diversity of permafrost response types to climate change [Shur & Jorgenson, 2007]. Thus, the regional permafrost thawing estimate does not yet have reliable information sources, unlike local forecasts, which are provided with a good variety of input data for modeling.

Answer 2: We are aware of the limitations brought by each indicator. The permafrost_cci indicators are so far the best available at a pan-Arctic scale to provide spatial and temporal estimates of the permafrost extent, ground temperature, and active layer depth. Abrupt thaw processes are considered in the "probability of thermokarst" indicator, which is also based on a dataset that estimates the probability of thermokarst to occur. 

Comment 3:The index "exposure" actually corresponds to the "danger", because not every event caused by permafrost thawing leads to the damage. For example, the development of thermokarst in the tundra is a danger, but only if thermokarst develops under a road or a building, then we can talk about the occurrence of exposure. The authors allow logical liberties, interpreting permafrost thawing as the exposure. The permafrost will not thaw everywhere and simultaneously, but only in certain landscapes. Not everywhere where permafrost is thawing the unfavorable processes become active. Not everywhere where dangerous processes develop, their magnitude can lead to damage to human infrastructure.

Answer 3: We adopt the following definitions of exposure as “the potential occurrence of a natural or human-induced physical event or trend that may cause loss of life, injury, or other health impacts, as well as damage and loss to property, infrastructure, livelihoods, service provision, ecosystems and environmental resources.” (L87). This means indeed that permafrost thaw becomes an exposure when it interacts with human activities. The method that we use in the APVI is to overlap spatially different indicators of exposure and adaptive capacity. To build the exposure index, we used the same method as for the whole vulnerability index, meaning that we scored each indicator and then for each area, we calculated the final score by averaging all exposure indicator scores. The final exposure index thus reflects the interactions between all these indicators, including the infrastructure indicator.  Thus, a place where permafrost thaws but where no one lives (no infrastructure) is not considered as exposed to permafrost thaw. 

Comment 4:"Exposure" and "Adaptive capacities" are functionally related when calculating the range of vulnerabilities (Figure 1). However, there are situations where "Low exposure" is adjacent to "Low capacity" or "Very high exposure" is adjacent to "Very high capacity".

Answer 4: Yes, this is correct. In some places, there might be low exposure to permafrost thaw, meaning that permafrost is not a big threat for populations, and meanwhile, this population has "low capacity" to answer to any change in permafrost. We understand that you suggest that it might not be necessary to look at the adaptive capacity in sub-regions that have a low exposure to permafrost thaw, it is a good suggestion. However, for consistency in the methods, we think it is necessary to apply the adaptive capacity to all sub-regions regardless of the level of exposure to permafrost thaw. This is because the data is averaged at the sub-regional scale due to the adaptive capacity indicators; thus even though a sub-region has a low exposure to permafrost thaw on average, there might still be areas within this sub-region where the exposure to permafrost thaw is higher.

Regarding sub-regions that have a very high exposure and adaptive capacity. This means that some regions are strongly affected by permafrost thaw but they also have very high capacity to answer to these changes, both economically and demographically, or because they have enough knowledge to address these issues and find and apply solutions.

When considering indices, authors are encouraged to discuss and analyze the following circumstances:

Comment 5:"Permafrost coverage" is not related to the danger to infrastructure. For example, continuous permafrost on a low ice content rocks is much less dangerous for construction than sporadic, rare, ice-bearing islands of permafrost on peatlands in valleys. Construction of a building on such peat island will lead to significant costs for engineering protection or a building repair. Even in areas with continuous permafrost the engineers choose the relatively favorable local areas. An analysis of spatial patterns of damage from permafrost processes showed its lack of dependence on the permafrost extent [Sergeev, Chesnokova, Morozova, 2015].

Answer 5: As for our answer for comment 3, both of the exposure and adaptive capacity indexes are obtained by aggregating and spatially overlapping different indicators. This means that areas in a continuous permafrost zone, with a low ice content, have a lower exposure score than areas in a continuous permafrost zone, with a high ice content.

Comment 6:"Change in permafrost extent" is not provided with reliable data in practice, since the modeling is related only to a highly simplified set of initial data "Permafrost_cci" and is poorly coordinated with "Ground ice coverage" from the Brown's Circumpolar Map. Soil warming trends are informative only if there are data on ice content, moisture regime of the active layer, evaporation and knowledge of the temperature distribution in the lower horizons of permafrost. If these data are not available, then simple averaging of warming trends over the territory will give incorrect results.

Answer 6: We are working on a pan-Arctic scale. While we cover a spatially large area, we have to accept the fact that data for these types of extent usually has lower spatial resolution and representation. The "Ground ice coverage" is an old product and regional or local datasets with better accuracy exist. However, we cannot use these products since they would introduce a bias in the analysis. We describe this in the limitation section of the manuscript.

Comment 7:Line 174. What the authors use should rather be designated as a "forecast of thermal subsidence", but not thermokarst. The development of thermokarst assumes a consistent and staged development of characteristic forms. Thermokarst coverage does not always correlate with the danger of thermal subsidence, since subsidence can occur, but thermokarst does not develop.

Answer 7: Thank you for this suggestion. This indicator is derived from a study published by Olefeld et al., 2016. We thus use the same terminology as they do to prevent any error in interpretation.

Comment 8:Line 162-165. Questionable conclusion. The highs ice-content leads to the slowing the permafrost thaws, but the final damage becomes greater. Such an index should only be used with a time scale in mind. The vulnerability index should be tied to the forecast period.

Answer 8: We agree with this. Most of the indicators that we use are static and reflect a situation at a certain point in time. We described the the temporality of the data in the method section and discussed this as a limitation is the limitation section (L443).

Comment 9:Line 225 The index appears to be incomplete and formal, since it lacks an indicator of differences in political management of adaptation in different countries. If in the Western Arctic the factor of self-government at the municipal level is significant, then in Russia adaptation is appointed by decree from Moscow and is carried out mainly with federal funds. This leads to different weights of the same indicators in different sectors of the Arctic. For example, half of the population of the gas field may have higher education, but there is no adaptation, because the boss did not order and did not allocate money.
The content and reasons for population migration in different places are also very different.
The economy in different countries is also structured differently. The income of the administrative territory may be high, but the budget structure may not allow for the development and implementation of adaptation plans.

Answer 9: the health indicator is used in the exposure index, thus just reflecting the level of exposure to anthrax due to permafrost thaw. It is not used to address adaptation capacities. There is a lack of a pan-Arctic health indicator, thus we could not use health as an indicator in the adaptive capacity index.

The adaptive capacity, by taking into account the knowledge, demographic, and economic capacities reflects the capacities in answering the exposure to permafrost thaw. Our vulnerability index should answer (not at local scale) the questions that you raise.

Regarding population migration patterns, we do not address this topic here as it is not the aim for this vulnerability index. A lot of good literature is available on this topic.

Regarding budgets: again, we are only pan-Arctic indicators and there are not available dataset on budget allocation in the Arctic sub-regions. As for the indicators of institutional capacity, these are limitations in this manuscript and we described them in the limitation section.

Comment 10:Line 477 Authors are advised to specify what the vulnerability index can be used for. It is said that it is basic, but basic for what? What classes of management decisions need it? The generalized vulnerability index seems to be of little use due to the mixing of assessments that are heterogeneous in time and at territories. It seems that a more productive path is to select management decisions for paired combinations of particular indices.

Answer 10: We inform the readers in the introduction about the audience that we target (L63-65). We agree, though, that since the index is pan-Arctic, it does not reflect local heterogeneities. Yet, it is a first step towards building a vulnerability index tailored to the permafrost context, using the limited but available dataset that exists at a pan-Arctic scale. We describe this in the limitation section of our manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

“The Arctic Permafrost Vulnerability Index” by J. Ramage et al. is a timely and interesting manuscript addressing the vulnerability of the Arctic’s permafrost. The authors proposed their solution. This work is among the first to so comprehensively address the vulnerabilities of the permafrost region. The complexities of vulnerability assessment are welcome, and this manuscript will strongly contribute to this area. I had questions regarding the assessment criteria. I won’t address that, in deference to the colossal work done by my colleagues. The uniqueness and timeliness of this work are recognized by me.
The manuscript structure has received some comments.
1)    The structure is not in balance. The manuscript’s core—descriptions of categories and assessment criteria—is presented in Section 2. Materials and Methods. I believe the Results section needs to present them. The submitted manuscript details a new methodology. This research yielded important results regarding categories and criteria.
2)    The Results section does not present the spatial analysis. For example, analyzing the data presented by the authors in the Supplementary, I found interesting spatial patterns in the distribution of vulnerability of the Permafrost Region.  If you add to this analysis, your work will be invaluable. It turns out that the Supplementary materials are more valuable than the manuscript itself.
I think this manuscript contributes new knowledge to the field. I have no doubt the scientific community will be very interested in these research results. However, I believe they could use some improvement in their present state. I believe the authors are capable of this.

Author Response

I would like to thank the reviewer for the encouraging words towards our work. This is appreciated. Following your comments and the ones from other reviewers, we have changed some parts of the manuscript. Please find the revised version attached.

Comment 1:    The structure is not in balance. The manuscript’s core—descriptions of categories and assessment criteria—is presented in Section 2. Materials and Methods. I believe the Results section needs to present them. The submitted manuscript details a new methodology. This research yielded important results regarding categories and criteria.

Answer 1: Indeed, our contribution also lies in the identification of suitable indicators. We therefore moved the section with definitions of indicators into the results section.

Comment 2:    The Results section does not present the spatial analysis. For example, analyzing the data presented by the authors in the Supplementary, I found interesting spatial patterns in the distribution of vulnerability of the Permafrost Region.  If you add to this analysis, your work will be invaluable. It turns out that the Supplementary materials are more valuable than the manuscript itself.

Answer 2: The data in the Supplementary is presented in the maps in Figure 3. To make this information clearer, we replaced the maps in Figure 3 by a graphic. We kept the maps in Supplementary.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript present a mapping of vulnerability to permafrost thaw-related risks based on a new qualitative index. The methodology is clearly presented, as well as the results. The work is an interesting contribution to the field of permafrost studies, meanwhile it is for now too limited in its scope. Two main things are missing : uncertainties assessment (or at least discussion) and climate change susceptibility.

The indicators used are subjected to uncertainties, at least some of them, such as numerical modelling derived permafrost coverage, change in permafrost extent, and ground ice coverage (content). These uncertainties should be discussed, for producing an additional map that gives spatial variability of the degree of confidence of the vulnerability map.

The second point is related to the fact that climate change projections (e.g., CMIP6 results) indicate spatially variable susceptibility to climate change that triggers permafrost thaw. For instance, depending on the considered place and the considered climate change scenario, permafrost disappearance may or may not be projected by 2050. This is an important information for permafrost thaw vulnerability that should be taken into account in its mapping.

So overall I recommend a major revision of this manuscript prior to its publication.

A few technical corrections below.

l 35-36 : “About five million people live on permafrost in the circumpolar north, among which three million live in settlements where permafrost will degrade and disappear by 2050 [1]” This sentence looks too affirmative to me. Projections are affected by uncertainties of various nature, and this should not be overlooked when using them.

L 77 : Figure 1 does not cover all possible combination of exposure / adaptation capacity (e.g.: to which vulnerability class belongs a site with high exposure and high adaptation capacity?)

l 117 : Table 1 lacks of readability.

Author Response

Comment 1: The indicators used are subjected to uncertainties, at least some of them, such as numerical modelling derived permafrost coverage, change in permafrost extent, and ground ice coverage (content). These uncertainties should be discussed, for producing an additional map that gives spatial variability of the degree of confidence of the vulnerability map.

Answer 1: We acknowledge the uncertainties present in our study but recognize that we did not communicate these clearly. To address this, we have added a paragraph in the limitations section of the manuscript specifically discussing the uncertainties related to the indicators. While we would have liked to assess the degree of confidence in the APVI, the sheer number of indicators involved makes this challenging. Additionally, many of these indicators do not provide their own estimates of uncertainty, such as ground ice content and coastal erosion, among others.

"Uncertainties related to indicators derived from models are a significant challenge in environmental and climate research. It is the case with many exposure indicators used in the APVI. These uncertainties arise from various sources, including model structure, input data, and parameter estimation. Structural uncertainties stem from the simplifications and assumptions made during model development, which can introduce biases and limit the model's ability to represent complex real-world systems. Input data uncertainties are due to the variability and potential inaccuracies in the data used to drive the models. This can include measurement errors, incomplete datasets, and temporal or spatial inconsistencies. Parameter uncertainties occur when the values used for model parameters are not well-constrained, leading to a range of possible outcomes. It is important to recognize that these uncertainties are inherent to the APVI."

Comment 2: The second point is related to the fact that climate change projections (e.g., CMIP6 results) indicate spatially variable susceptibility to climate change that triggers permafrost thaw. For instance, depending on the considered place and the considered climate change scenario, permafrost disappearance may or may not be projected by 2050. This is an important information for permafrost thaw vulnerability that should be taken into account in its mapping.

Answer 2: This is a very good point. Indeed, the APVI does not draw any future scenario but just presents the current vulnerability status for the subregions. To clarify this, we added some text in the limitation section of the manuscript:

"Many APVI indicators are static and do not consider recent or future changes that reflect trends in the physical and socio-economic environments. Only one set of selected indicators brings a temporal aspect to express the exposure to permafrost thaw. None of the selected indicators consider future changes. In this regard, the APVI does not evaluate or provide scenarios regarding future vulnerability to permafrost thaw but reflects the current vulnerability to permafrost thaw."

 

l 35-36 : “About five million people live on permafrost in the circumpolar north, among which three million live in settlements where permafrost will degrade and disappear by 2050 [1]” This sentence looks too affirmative to me. Projections are affected by uncertainties of various nature, and this should not be overlooked when using them.

We modified the sentence to "Approximately five million people live on permafrost in the circumpolar north, with estimates indicating that three million live in areas where permafrost may degrade or disappear by 2050. [1]"

L 77 : Figure 1 does not cover all possible combination of exposure / adaptation capacity (e.g.: to which vulnerability class belongs a site with high exposure and high adaptation capacity?)

We adjusted the Figure to represent all possible combinations.

l 117 : Table 1 lacks of readability.

We adjusted the font for the table so that it is easier to read. We think that the journal will have a special format for the Tables.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The additional discussions introduced in this revised version give the relevant information and thoughts for a relevant use of the produced scientific material. With these improvements the manuscript is in a good enough shape for publication.