the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Nov 2024
Quantifying economic risks to dairy farms from volcanic hazards in Taranaki, New Zealand
Abstract. The volcanic hazard and risk science for Taranaki Mounga (Taranaki volcano) in New Zealand is in an advanced state, with robust probabilistic data and a series of direct impact scenarios modelled for the region. Here we progress this work and demonstrate a method to provide risk information that is nuanced for factors such as location and economic sector, and considers the dynamic nature of volcanism with hazards potentially repeated over time. Recognising the fundamental importance of the dairy sector to Taranaki Region, this paper provides valuable insights on the potential impacts and risks to heterogeneous dairy cattle farms within the region from volcanic hazards. We provide volcanic impact and risk metrics in economic or monetary terms in order to improve its relevance to decision makers while reducing the complexity of the impacts. To do this, we developed a dynamic, multi-event farm system model of response and recovery, which takes in hazard intensity metrics from a series of volcanic events, and generates the resulting annualised revenues, expenditures and recovery costs through time. The model is formulated in a generalised way such that it can be used for various other hazard types and agricultural land uses. In our application of the model, we create and apply a suite of ten thousand simulations that capture different iterations of possible future volcanic activity over a 50-year period. These include the generation of lahars following eruptions and associated failures for transport and water supply networks. Farms at five case study locations were modelled, to capture the diversity in farm management and the spatial variation in hazard intensities and likelihoods across the region. We provide summaries of the distributions of economic impacts generated, both for individual events and for the 50-year volcanic future horizon. Drawing the information together, we also summarise the results for each case study farm in terms of the Value at Risk statistic. For the case study farms with negligible lahar risk we find, with 90 % confidence, that volcanic losses over the next 50 years will not exceed around 10 % of property value. By comparison, for the farm with the most severe lahar and ashfall exposure, we find that at the same level of confidence, losses extend to approximately half the property value. These results indicate that with access to sufficient risk information, we should anticipate volcanic risk as playing an important role in shaping the future dairy sector in Taranaki Region. The modelling pipeline and assessment metrics demonstrated in this paper could be used to assess mitigation and adaptation strategies to reduce the risk from volcanic hazards and improve the resilience of farm businesses.
- Preprint
(2476 KB) - Metadata XML
-
Supplement
(14 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-3619', Anonymous Referee #1, 28 Dec 2024
The manuscript presents a clear, well structured, and fullsome description of a novel model built to estimate the economic impacts of volcanic hazards on dairy farms. The research artfully combines physical hazard impact data, infrastructure system disruption information and farm-level impact, response and recovery.
The manuscript is of excellent quality and my comments are limited to a request for some more consideration of limitations and/or future refinements, as well as some small technical corrections.
Specific comments:
- Agricultural activities are seasonal and the impact of an event (and the likely management decisions made) will likely depend on the time of year that the hazard occurs. This does not seem to be discussed in the paper. This needs to be discussed as a limitation or opportunity for future research at the very least.
- It appears that farm impacts (and subsequent recovery) are based on ‘actual’ hazard impacts and are reactive. Were pre-emptive adaptations e.g. relocation of stock ever considered? Perhaps add to discussion around line 500 (so to include not just reactive but pro-active actions when warnings are issued).
- It was not clear how much, if at all, post-event commodity price changes are considered (both impact and recovery costs)? Increased transportation costs were discussed but not commodity prices themselves. It might be helpful to comment on this.
Technical corrections
FIGURE 1 – include definitions provided in text (i,j) (such that the figure can be interpretted in isolation)
Line 89 – states Figure 1 shows that business operation and behaviour needs to be included. Not immediately clear which variable (s) this relates to. Consider being more explicit in the text (which variable in Figure 1 are you refering to?).
Line 103 – consider a semi colon to separate the two examples
Line 191 – the j,e+1 should be a subscript
Line 423 – extra ‘any’
Line 494 – should be lose not loose
Citation: https://doi.org/10.5194/egusphere-2024-3619-RC1 -
AC1: 'Reply on RC1', Nicola McDonald, 22 Jan 2025
Response to Referee 1
We are thankful for the time Referee 1 has taken to review the manuscript. The manuscript documents a significant piece of work for our team involving many contributors, each bringing different knowledge and disciplinary backgrounds. We are thus very pleased to have received the positive comments from Referee 1.
We respond to the Referee’s specific comments below:
Specific Comments
Comment 1: Agricultural activities are seasonal and the impact of an event (and the likely management decisions made) will likely depend on the time of year that the hazard occurs. This does not seem to be discussed in the paper. This needs to be discussed as a limitation or opportunity for future research at the very least.
We agree that seasonality is an important matter to consider for agriculture. Some aspects of seasonality were considered in the modelling (there is a lot of detail contained in the manuscript so it might be easy to miss these details):
- The structure of the vulnerability functions for ashfall and transport allow for a period of lower vulnerability to be recognised when calculating the damage state. For the transport damage state calculations, a period of lower vulnerability was applied to days each year outside of the milking period (see line 310 and Appendix A2). For the ashfall damage functions, however, we decided not to apply the Craig et al (2021) adjustment mechanism to account for a period of the year with lower vulnerability to ashfall (see footnote 6). Our reasoning was that these farms are generally highly balanced systems and even though there may be periods of the year where activity is reduced (e.g. no milking), ashfall would likely still be disruptive year-round. This could be a matter for further consideration in future studies which can be noted in the text.
- The function that determines impact duration for impact states where inaccessibility/transport is the dominant impact take into consideration the remaining portion of the year that is the milking season (this is noted at line 380 but the explanation could be expanded).
We propose adding a comment in the conclusion that highlights the importance of seasonality to agriculture and recommending that in any future applications careful consideration needs to be given to the way in which impacts will vary depending on season.
Comment 2: It appears that farm impacts (and subsequent recovery) are based on 'actual' hazard impacts and are reactive. Were pre-emptive adaptations e.g. relocation of stock ever considered? Perhaps add to discussion around line 500 (so to include not just reactive but pro-active actions when warnings are issued).
One point to note is that the assignment of ashfall damage states is based on the study by Craig et al (2021). To the extent that the farms surveyed in that study undertook pre-emptive actions, we will have given some account to pre-emptive measures in the calculations of the impacts.
In our opinion the two topics most worthy of consideration for improving the impact calculation in relation to pre-emptive measures relate to actions taken to preserve human life and health, ie.
- Evacuations during and prior to an eruption when volcanic unrest is high
- Evacuations when lahar risk is considered high
Although many farmers may wish to evacuate livestock, earlier research has indicated that given the numbers of stock involved and the constraints on resources necessary for moving stock, it is simply unlikely that stock can be moved (see Wilson et al. 2009. Modelling livestock evacuation following a volcanic eruption: An example from Taranaki volcano, New Zealand. New Zealand Journal of Agricultural Research,52(1), 99–110. https://doi.org/10.1080/0028823090951049).
There is uncertainty around the governance decisions that would be put in place to guide human evacuations, and the extent to which people will comply with such evacuations. Appetite for evacuation is also likely to change over time as risk tolerance is also dynamic and path dependent (i.e. influenced by recent experiences). Hence recent work indicates that evacuation requirements might be quite dynamic (see Coultas, K. Identifying Evacuation Planning Considerations for Complex Volcanic Crises: A Case Study from Taranaki, Aotearoa New Zealand. Masters Thesis. University of Canterbury). For now, we have not attempted to include evacuations in the consideration of impacts. We also believe that further research is necessary in terms of lahar hazard modelling and risk assessment to be able to sufficiently include lahar evacuations in the impact assessment. A brief mention is already made in the manuscript at line 468 on this topic.
Overall, we can be clearer in the text (around line 500 and elsewhere including in the conclusion) that in-depth consideration of pre-emptive measures is a limitation of the study and worthy of future research.
We would also like to mention that we intend to undertake a further study that examines the risk implications of alternative land uses (i.e. a subset of potential pre-emptive measures).
Comment 3: It was not clear how much, if at all, post-event commodity price changes are considered (both impact and recovery costs)? Increased transportation costs were discussed but not commodity prices themselves. It might be helpful to comment on this.
For the most part we did not consider commodity price changes in determining the impacts. The major exception was in terms of situations where farms must sell stock due to feed shortages but are still able to get their stock to market (ie not subject to major transportation disruption). In these situations, it was assumed that the price realised for stock sales is only 50% of the normal stock value due to the ‘desperate’ situation of farmers and to some extent the flooding of stock in local markets at the time of sale. There is a reference to this around line 420 of the manuscript. With regards to other commodities produced by farms (raw milk) we did not assume any commodity price changes. This is justified given that most New Zealand’s dairy products are for export, and New Zealand is generally considered a price taker in these markets. It is possible that we could have further considered price changes for commodity inputs to farms (e.g. feed and transport) and this would have caused some increase in the impacts calculated. We can make a comment in the manuscript on this limitation.
Conceptually the model can include price changes when calculating impacts. For example, given that commodity expenditure line items of Fi,j are the product of both commodity quantity and commodity price, the impact scalars applied to these (S0j) could account for both quantity and price effects. We can expand the examples of S0j given around line 104 of the manuscript to make it clearer to the reader that price effects can be considered.
Technical Corrections
These are detailed in the Referee’s original post. We thank the referee for identifying these and we can make all the necessary corrections to the manuscript.
Citation: https://doi.org/10.5194/egusphere-2024-3619-AC1
-
RC2: 'Comment on egusphere-2024-3619', Anonymous Referee #2, 06 Jan 2025
The paper is an excellent case study of a risk modelling approach for hazards affecting agriculture. The paper is very well written and structured, and the results for impacts on dairy are well considered and reasoned. It is my opinion that the paper would be of high value and keen interest to the readership of the journal, and recommend publishing subject to the minor revisions below.
Certain aspects were not immediately clear and some limited explanation or expansion would be appreciated:
1) the consideration of dangers for a dairy operation were detailed however no consideration for potential injury or death of workers or operators seems to have been considered. In particular, would farms simulated to be affected directly by lahars (or pycroclastic flows as proxied by lahars in the model) have any associated risk to humans. Similarly, would ash fall have any consequences for human health? Injury and/or loss of life would be tragic, costly and detrimental for returning a farm back to full operating capacity. The conclusion of the report describes the findings as informative for farmers and decision-makers in understanding the risks to dairy farming. I would consider any potential impact on human health or life to be key information for farmers in this area of inquiry.
Some explanation of why these elements were not considered, why the volcanic impacts are not considered to pose a risk to farm workers or operators, or acknowledging any limitations in the method for considering these elements would strengthen the paper.
2) Similarly, in the cases of 100% destocking with the event of lahars, is the assumption that this is due to animal mortality? If so, does the model consider no recovery value for those animals (where in other cases some recovery can be achieved by selling stock)? Clarifying this would be appreciated.
3) In the case of recovery prices for the replacement of farm assets, do these costs include the demolition and/or clearing costs for removing the assets they are replacing. Clarification or acknowledgement of any limitations in this area would be appreciated.
4) The explanation of why impact duration (or recovery time) was used as the proxy for severity in the event of overlapping events could be expanded. Are there cases of longer term but minor impacts or conversely short term but high impacts? Would it be possible to instead, use average impact over the remaining years as a proxy of severity? Slightly more explanation or acknowledgement of the minor limitation would be appreciated.
5) I had difficulty in understanding how I was meant to interpret Figure A1b in terms of the relationship between the two axes and which measure was determining what. Some additional clarification on how to read this graph would be appreciated.
6) No source was provided for the data in Figure 4e.
7) Why are the total recovery costs spread evenly over the impact duration. Would traditional farm assets (i.e. dairy platform, shed, house) not have a standard loan and repayment schedule or are these considered in the impact duration?There were some minor typos that should be corrected prior to publication:
Pg. 7, line 162: a double '>' is used, unclear whether this was intentional as a 'much' greater than.
Pg. 11, line 231: the sentence ending "...in this case" is missing a full-stop.
Pg, 12, description of Figure 4.: there is a double semi-colon following the phrase "...and selective rivers for the study".
Pg. 16, line 302: "ie" should be "i.e."
Pg. 22, line 470-471: "...Box and Whisker plots" does not require capitalisation
Pg. 24, line 494: "loose" should be "lose"
Pg. 25, description of Figure 9.: double "of"
Pg. 36, line 728: "See Table ??..." needs correct reference to table
I do not think any further action is required on the following point, but might be points of consideration for the authors:
1) In the final economic impact presented in section 6.2.1, NPVs are presented for the different scenarios over the 50-year period. I was unsure of the logic of using NPVs over these periods when volcanic events were spread dependent on the modelled volcanic events. From a policy or decision makers perspective this makes sense in calculating risk over the next 50-years. However the context of the paper seems more directed to individual farm and thus discounting from individual events throughout their recovery (and calculating mean impacts over however many events they have during the 50-year period), may have been a more relevant metric for farmers.Citation: https://doi.org/10.5194/egusphere-2024-3619-RC2 -
AC2: 'Reply on RC2', Nicola McDonald, 22 Jan 2025
Response to Referee 2
We are very appreciative of the time you have taken to read the manuscript and provide comments. The comments are useful and will help us to improve the manuscript.
Our responses are provided below.
Specific Comments
[Comment 1] the consideration of dangers for a dairy operation were detailed however no consideration for potential injury or death of workers or operators seems to have been considered. In particular, would farms simulated to be affected directly by lahars (or pycroclastic flows as proxied by lahars in the model) have any associated risk to humans. Similarly, would ash fall have any consequences for human health? Injury and/or loss of life would be tragic, costly and detrimental for returning a farm back to full operating capacity. The conclusion of the report describes the findings as informative for farmers and decision-makers in understanding the risks to dairy farming. I would consider any potential impact on human health or life to be key information for farmers in this area of inquiry.
Some explanation of why these elements were not considered, why the volcanic impacts are not considered to pose a risk to farm workers or operators, or acknowledging any limitations in the method for considering these elements would strengthen the paper.
We agree that risks to human life and health, including for farmers, are important considerations in the context of a future Taranaki eruption. In terms of economic impacts, we believe that it will be actions taken to reduce these risks (e.g. farm evacuations) that will be particularly important to consider. We consider this comment as overlapping with Comment 2 of Reviewer 1. Please therefore also refer to our responses to Reviewer 1.
[Comment 2] Similarly, in the cases of 100% destocking with the event of lahars, is the assumption that this is due to animal mortality? If so, does the model consider no recovery value for those animals (where in other cases some recovery can be achieved by selling stock)? Clarifying this would be appreciated.
We agree that more information should be provided to clarify the parameters used in the modelling which determine recovery value for stock. Where the less severe transport damage states are experienced, it is assumed that a proportion of the stock that cannot be supported on site (due to loss of feed from ashfall damage and/or loss of water supply) can be transported and sold, with 50% of normal stock value realised. We can communicate this by adding additional information to the Appendix and a small change to the main body of the manuscript. In the case of experiencing a lahar, it is possible that animal mortality will not initially be 100%. Nevertheless, it was considered appropriate to assume no recovery of stock value under IS4 for lahars given the likely concurrent impacts on the transport network preventing easy transportation of stock for sale.
[Comment 3] In the case of recovery prices for the replacement of farm assets, do these costs include the demolition and/or clearing costs for removing the assets they are replacing. Clarification or acknowledgement of any limitations in this area would be appreciated.
We have not included asset removal costs. We can add a comment to the manuscript to note this as a limitation.
[Comment 4] The explanation of why impact duration (or recovery time) was used as the proxy for severity in the event of overlapping events could be expanded. Are there cases of longer term but minor impacts or conversely short term but high impacts? Would it be possible to instead, use average impact over the remaining years as a proxy of severity? Slightly more explanation or acknowledgement of the minor limitation would be appreciated.
This proxy was chosen in part to reflect ease of implementation. We are in the process of refactoring the model to implement it as a more flexible R package so that it can be applied for other studies including for other land uses. We thank the reviewer for this useful suggestion and will investigate doing this as we continue to develop the model.
[Comment 5] I had difficulty in understanding how I was meant to interpret Figure A1b in terms of the relationship between the two axes and which measure was determining what. Some additional clarification on how to read this graph would be appreciated.
Agree that the figure could be better explained. We can make the corrections to the Figure in the Appendix.
[Comment 6] No source was provided for the data in Figure 4e.
Thank you for identifying this. We can include this in the caption.
[Comment 7] Why are the total recovery costs spread evenly over the impact duration. Would traditional farm assets (i.e. dairy platform, shed, house) not have a standard loan and repayment schedule or are these considered in the impact duration?
Recovery costs are spread out evenly over the impact duration as this was considered the most appropriately way to generically model a range of asset types with varying timeframes at which payments for recovery and replacement may occur (e.g. immediately following the event, following completion of land rehabilitation, incrementally over time). We can add a comment around line 125 to make it more clear to the reader that this is a simplifying assumption implemented in the model.
Although payments for replacement of larger assets may be able to be financed through loans, we based the calculations on the perspective of when the payments to suppliers (e.g., shed builders) are due. We also note that when producing results as discounted cash flows (as we have undertaken to do), there should be little difference between recording the full payment when it is due, versus recording the schedule of loan payments over time with interest included.
Typos and Corrections
Thank you for identifying these errors. We will make the necessary corrections in the manuscript.
Further point
[Comment 1] I do not think any further action is required on the following point, but might be points of consideration for the authors:
In the final economic impact presented in section 6.2.1, NPVs are presented for the different scenarios over the 50-year period. I was unsure of the logic of using NPVs over these periods when volcanic events were spread dependent on the modelled volcanic events. From a policy or decision makers perspective this makes sense in calculating risk over the next 50-years. However the context of the paper seems more directed to individual farm and thus discounting from individual events throughout their recovery (and calculating mean impacts over however many events they have during the 50-year period), may have been a more relevant metric for farmers.
This is an interesting observation. When writing the Results section, we compiled a long list of potential ways we could put together information from the modelling and alternative graphs and figures that could be produced. We recognised that different metrics and graphs emphasised different information, but also that we could only select a few metrics and graphs to include. Overall, we believe that we have the balance right.
We preferred to keep with discounted information in Section 6.2.1. The reason being that ‘discounted’ cash flows are a very common metric to agricultural stakeholders (farms, local council etc). For example, they are commonly used to compare alternative land use options, or the implications of policy options. We also note that ‘future earnings’ is one way to estimate a farms value, and this uses discounted cash flows over time. But to date, we have had little understanding of how these values are at risk of being achieved given future volcanic activity. Producing the results as we have done will thus be very insightful for those wanting to better take into consideration volcanic risk in understanding the economic value of farms.
Citation: https://doi.org/10.5194/egusphere-2024-3619-AC2
-
AC2: 'Reply on RC2', Nicola McDonald, 22 Jan 2025
Status: closed
-
RC1: 'Comment on egusphere-2024-3619', Anonymous Referee #1, 28 Dec 2024
The manuscript presents a clear, well structured, and fullsome description of a novel model built to estimate the economic impacts of volcanic hazards on dairy farms. The research artfully combines physical hazard impact data, infrastructure system disruption information and farm-level impact, response and recovery.
The manuscript is of excellent quality and my comments are limited to a request for some more consideration of limitations and/or future refinements, as well as some small technical corrections.
Specific comments:
- Agricultural activities are seasonal and the impact of an event (and the likely management decisions made) will likely depend on the time of year that the hazard occurs. This does not seem to be discussed in the paper. This needs to be discussed as a limitation or opportunity for future research at the very least.
- It appears that farm impacts (and subsequent recovery) are based on ‘actual’ hazard impacts and are reactive. Were pre-emptive adaptations e.g. relocation of stock ever considered? Perhaps add to discussion around line 500 (so to include not just reactive but pro-active actions when warnings are issued).
- It was not clear how much, if at all, post-event commodity price changes are considered (both impact and recovery costs)? Increased transportation costs were discussed but not commodity prices themselves. It might be helpful to comment on this.
Technical corrections
FIGURE 1 – include definitions provided in text (i,j) (such that the figure can be interpretted in isolation)
Line 89 – states Figure 1 shows that business operation and behaviour needs to be included. Not immediately clear which variable (s) this relates to. Consider being more explicit in the text (which variable in Figure 1 are you refering to?).
Line 103 – consider a semi colon to separate the two examples
Line 191 – the j,e+1 should be a subscript
Line 423 – extra ‘any’
Line 494 – should be lose not loose
Citation: https://doi.org/10.5194/egusphere-2024-3619-RC1 -
AC1: 'Reply on RC1', Nicola McDonald, 22 Jan 2025
Response to Referee 1
We are thankful for the time Referee 1 has taken to review the manuscript. The manuscript documents a significant piece of work for our team involving many contributors, each bringing different knowledge and disciplinary backgrounds. We are thus very pleased to have received the positive comments from Referee 1.
We respond to the Referee’s specific comments below:
Specific Comments
Comment 1: Agricultural activities are seasonal and the impact of an event (and the likely management decisions made) will likely depend on the time of year that the hazard occurs. This does not seem to be discussed in the paper. This needs to be discussed as a limitation or opportunity for future research at the very least.
We agree that seasonality is an important matter to consider for agriculture. Some aspects of seasonality were considered in the modelling (there is a lot of detail contained in the manuscript so it might be easy to miss these details):
- The structure of the vulnerability functions for ashfall and transport allow for a period of lower vulnerability to be recognised when calculating the damage state. For the transport damage state calculations, a period of lower vulnerability was applied to days each year outside of the milking period (see line 310 and Appendix A2). For the ashfall damage functions, however, we decided not to apply the Craig et al (2021) adjustment mechanism to account for a period of the year with lower vulnerability to ashfall (see footnote 6). Our reasoning was that these farms are generally highly balanced systems and even though there may be periods of the year where activity is reduced (e.g. no milking), ashfall would likely still be disruptive year-round. This could be a matter for further consideration in future studies which can be noted in the text.
- The function that determines impact duration for impact states where inaccessibility/transport is the dominant impact take into consideration the remaining portion of the year that is the milking season (this is noted at line 380 but the explanation could be expanded).
We propose adding a comment in the conclusion that highlights the importance of seasonality to agriculture and recommending that in any future applications careful consideration needs to be given to the way in which impacts will vary depending on season.
Comment 2: It appears that farm impacts (and subsequent recovery) are based on 'actual' hazard impacts and are reactive. Were pre-emptive adaptations e.g. relocation of stock ever considered? Perhaps add to discussion around line 500 (so to include not just reactive but pro-active actions when warnings are issued).
One point to note is that the assignment of ashfall damage states is based on the study by Craig et al (2021). To the extent that the farms surveyed in that study undertook pre-emptive actions, we will have given some account to pre-emptive measures in the calculations of the impacts.
In our opinion the two topics most worthy of consideration for improving the impact calculation in relation to pre-emptive measures relate to actions taken to preserve human life and health, ie.
- Evacuations during and prior to an eruption when volcanic unrest is high
- Evacuations when lahar risk is considered high
Although many farmers may wish to evacuate livestock, earlier research has indicated that given the numbers of stock involved and the constraints on resources necessary for moving stock, it is simply unlikely that stock can be moved (see Wilson et al. 2009. Modelling livestock evacuation following a volcanic eruption: An example from Taranaki volcano, New Zealand. New Zealand Journal of Agricultural Research,52(1), 99–110. https://doi.org/10.1080/0028823090951049).
There is uncertainty around the governance decisions that would be put in place to guide human evacuations, and the extent to which people will comply with such evacuations. Appetite for evacuation is also likely to change over time as risk tolerance is also dynamic and path dependent (i.e. influenced by recent experiences). Hence recent work indicates that evacuation requirements might be quite dynamic (see Coultas, K. Identifying Evacuation Planning Considerations for Complex Volcanic Crises: A Case Study from Taranaki, Aotearoa New Zealand. Masters Thesis. University of Canterbury). For now, we have not attempted to include evacuations in the consideration of impacts. We also believe that further research is necessary in terms of lahar hazard modelling and risk assessment to be able to sufficiently include lahar evacuations in the impact assessment. A brief mention is already made in the manuscript at line 468 on this topic.
Overall, we can be clearer in the text (around line 500 and elsewhere including in the conclusion) that in-depth consideration of pre-emptive measures is a limitation of the study and worthy of future research.
We would also like to mention that we intend to undertake a further study that examines the risk implications of alternative land uses (i.e. a subset of potential pre-emptive measures).
Comment 3: It was not clear how much, if at all, post-event commodity price changes are considered (both impact and recovery costs)? Increased transportation costs were discussed but not commodity prices themselves. It might be helpful to comment on this.
For the most part we did not consider commodity price changes in determining the impacts. The major exception was in terms of situations where farms must sell stock due to feed shortages but are still able to get their stock to market (ie not subject to major transportation disruption). In these situations, it was assumed that the price realised for stock sales is only 50% of the normal stock value due to the ‘desperate’ situation of farmers and to some extent the flooding of stock in local markets at the time of sale. There is a reference to this around line 420 of the manuscript. With regards to other commodities produced by farms (raw milk) we did not assume any commodity price changes. This is justified given that most New Zealand’s dairy products are for export, and New Zealand is generally considered a price taker in these markets. It is possible that we could have further considered price changes for commodity inputs to farms (e.g. feed and transport) and this would have caused some increase in the impacts calculated. We can make a comment in the manuscript on this limitation.
Conceptually the model can include price changes when calculating impacts. For example, given that commodity expenditure line items of Fi,j are the product of both commodity quantity and commodity price, the impact scalars applied to these (S0j) could account for both quantity and price effects. We can expand the examples of S0j given around line 104 of the manuscript to make it clearer to the reader that price effects can be considered.
Technical Corrections
These are detailed in the Referee’s original post. We thank the referee for identifying these and we can make all the necessary corrections to the manuscript.
Citation: https://doi.org/10.5194/egusphere-2024-3619-AC1
-
RC2: 'Comment on egusphere-2024-3619', Anonymous Referee #2, 06 Jan 2025
The paper is an excellent case study of a risk modelling approach for hazards affecting agriculture. The paper is very well written and structured, and the results for impacts on dairy are well considered and reasoned. It is my opinion that the paper would be of high value and keen interest to the readership of the journal, and recommend publishing subject to the minor revisions below.
Certain aspects were not immediately clear and some limited explanation or expansion would be appreciated:
1) the consideration of dangers for a dairy operation were detailed however no consideration for potential injury or death of workers or operators seems to have been considered. In particular, would farms simulated to be affected directly by lahars (or pycroclastic flows as proxied by lahars in the model) have any associated risk to humans. Similarly, would ash fall have any consequences for human health? Injury and/or loss of life would be tragic, costly and detrimental for returning a farm back to full operating capacity. The conclusion of the report describes the findings as informative for farmers and decision-makers in understanding the risks to dairy farming. I would consider any potential impact on human health or life to be key information for farmers in this area of inquiry.
Some explanation of why these elements were not considered, why the volcanic impacts are not considered to pose a risk to farm workers or operators, or acknowledging any limitations in the method for considering these elements would strengthen the paper.
2) Similarly, in the cases of 100% destocking with the event of lahars, is the assumption that this is due to animal mortality? If so, does the model consider no recovery value for those animals (where in other cases some recovery can be achieved by selling stock)? Clarifying this would be appreciated.
3) In the case of recovery prices for the replacement of farm assets, do these costs include the demolition and/or clearing costs for removing the assets they are replacing. Clarification or acknowledgement of any limitations in this area would be appreciated.
4) The explanation of why impact duration (or recovery time) was used as the proxy for severity in the event of overlapping events could be expanded. Are there cases of longer term but minor impacts or conversely short term but high impacts? Would it be possible to instead, use average impact over the remaining years as a proxy of severity? Slightly more explanation or acknowledgement of the minor limitation would be appreciated.
5) I had difficulty in understanding how I was meant to interpret Figure A1b in terms of the relationship between the two axes and which measure was determining what. Some additional clarification on how to read this graph would be appreciated.
6) No source was provided for the data in Figure 4e.
7) Why are the total recovery costs spread evenly over the impact duration. Would traditional farm assets (i.e. dairy platform, shed, house) not have a standard loan and repayment schedule or are these considered in the impact duration?There were some minor typos that should be corrected prior to publication:
Pg. 7, line 162: a double '>' is used, unclear whether this was intentional as a 'much' greater than.
Pg. 11, line 231: the sentence ending "...in this case" is missing a full-stop.
Pg, 12, description of Figure 4.: there is a double semi-colon following the phrase "...and selective rivers for the study".
Pg. 16, line 302: "ie" should be "i.e."
Pg. 22, line 470-471: "...Box and Whisker plots" does not require capitalisation
Pg. 24, line 494: "loose" should be "lose"
Pg. 25, description of Figure 9.: double "of"
Pg. 36, line 728: "See Table ??..." needs correct reference to table
I do not think any further action is required on the following point, but might be points of consideration for the authors:
1) In the final economic impact presented in section 6.2.1, NPVs are presented for the different scenarios over the 50-year period. I was unsure of the logic of using NPVs over these periods when volcanic events were spread dependent on the modelled volcanic events. From a policy or decision makers perspective this makes sense in calculating risk over the next 50-years. However the context of the paper seems more directed to individual farm and thus discounting from individual events throughout their recovery (and calculating mean impacts over however many events they have during the 50-year period), may have been a more relevant metric for farmers.Citation: https://doi.org/10.5194/egusphere-2024-3619-RC2 -
AC2: 'Reply on RC2', Nicola McDonald, 22 Jan 2025
Response to Referee 2
We are very appreciative of the time you have taken to read the manuscript and provide comments. The comments are useful and will help us to improve the manuscript.
Our responses are provided below.
Specific Comments
[Comment 1] the consideration of dangers for a dairy operation were detailed however no consideration for potential injury or death of workers or operators seems to have been considered. In particular, would farms simulated to be affected directly by lahars (or pycroclastic flows as proxied by lahars in the model) have any associated risk to humans. Similarly, would ash fall have any consequences for human health? Injury and/or loss of life would be tragic, costly and detrimental for returning a farm back to full operating capacity. The conclusion of the report describes the findings as informative for farmers and decision-makers in understanding the risks to dairy farming. I would consider any potential impact on human health or life to be key information for farmers in this area of inquiry.
Some explanation of why these elements were not considered, why the volcanic impacts are not considered to pose a risk to farm workers or operators, or acknowledging any limitations in the method for considering these elements would strengthen the paper.
We agree that risks to human life and health, including for farmers, are important considerations in the context of a future Taranaki eruption. In terms of economic impacts, we believe that it will be actions taken to reduce these risks (e.g. farm evacuations) that will be particularly important to consider. We consider this comment as overlapping with Comment 2 of Reviewer 1. Please therefore also refer to our responses to Reviewer 1.
[Comment 2] Similarly, in the cases of 100% destocking with the event of lahars, is the assumption that this is due to animal mortality? If so, does the model consider no recovery value for those animals (where in other cases some recovery can be achieved by selling stock)? Clarifying this would be appreciated.
We agree that more information should be provided to clarify the parameters used in the modelling which determine recovery value for stock. Where the less severe transport damage states are experienced, it is assumed that a proportion of the stock that cannot be supported on site (due to loss of feed from ashfall damage and/or loss of water supply) can be transported and sold, with 50% of normal stock value realised. We can communicate this by adding additional information to the Appendix and a small change to the main body of the manuscript. In the case of experiencing a lahar, it is possible that animal mortality will not initially be 100%. Nevertheless, it was considered appropriate to assume no recovery of stock value under IS4 for lahars given the likely concurrent impacts on the transport network preventing easy transportation of stock for sale.
[Comment 3] In the case of recovery prices for the replacement of farm assets, do these costs include the demolition and/or clearing costs for removing the assets they are replacing. Clarification or acknowledgement of any limitations in this area would be appreciated.
We have not included asset removal costs. We can add a comment to the manuscript to note this as a limitation.
[Comment 4] The explanation of why impact duration (or recovery time) was used as the proxy for severity in the event of overlapping events could be expanded. Are there cases of longer term but minor impacts or conversely short term but high impacts? Would it be possible to instead, use average impact over the remaining years as a proxy of severity? Slightly more explanation or acknowledgement of the minor limitation would be appreciated.
This proxy was chosen in part to reflect ease of implementation. We are in the process of refactoring the model to implement it as a more flexible R package so that it can be applied for other studies including for other land uses. We thank the reviewer for this useful suggestion and will investigate doing this as we continue to develop the model.
[Comment 5] I had difficulty in understanding how I was meant to interpret Figure A1b in terms of the relationship between the two axes and which measure was determining what. Some additional clarification on how to read this graph would be appreciated.
Agree that the figure could be better explained. We can make the corrections to the Figure in the Appendix.
[Comment 6] No source was provided for the data in Figure 4e.
Thank you for identifying this. We can include this in the caption.
[Comment 7] Why are the total recovery costs spread evenly over the impact duration. Would traditional farm assets (i.e. dairy platform, shed, house) not have a standard loan and repayment schedule or are these considered in the impact duration?
Recovery costs are spread out evenly over the impact duration as this was considered the most appropriately way to generically model a range of asset types with varying timeframes at which payments for recovery and replacement may occur (e.g. immediately following the event, following completion of land rehabilitation, incrementally over time). We can add a comment around line 125 to make it more clear to the reader that this is a simplifying assumption implemented in the model.
Although payments for replacement of larger assets may be able to be financed through loans, we based the calculations on the perspective of when the payments to suppliers (e.g., shed builders) are due. We also note that when producing results as discounted cash flows (as we have undertaken to do), there should be little difference between recording the full payment when it is due, versus recording the schedule of loan payments over time with interest included.
Typos and Corrections
Thank you for identifying these errors. We will make the necessary corrections in the manuscript.
Further point
[Comment 1] I do not think any further action is required on the following point, but might be points of consideration for the authors:
In the final economic impact presented in section 6.2.1, NPVs are presented for the different scenarios over the 50-year period. I was unsure of the logic of using NPVs over these periods when volcanic events were spread dependent on the modelled volcanic events. From a policy or decision makers perspective this makes sense in calculating risk over the next 50-years. However the context of the paper seems more directed to individual farm and thus discounting from individual events throughout their recovery (and calculating mean impacts over however many events they have during the 50-year period), may have been a more relevant metric for farmers.
This is an interesting observation. When writing the Results section, we compiled a long list of potential ways we could put together information from the modelling and alternative graphs and figures that could be produced. We recognised that different metrics and graphs emphasised different information, but also that we could only select a few metrics and graphs to include. Overall, we believe that we have the balance right.
We preferred to keep with discounted information in Section 6.2.1. The reason being that ‘discounted’ cash flows are a very common metric to agricultural stakeholders (farms, local council etc). For example, they are commonly used to compare alternative land use options, or the implications of policy options. We also note that ‘future earnings’ is one way to estimate a farms value, and this uses discounted cash flows over time. But to date, we have had little understanding of how these values are at risk of being achieved given future volcanic activity. Producing the results as we have done will thus be very insightful for those wanting to better take into consideration volcanic risk in understanding the economic value of farms.
Citation: https://doi.org/10.5194/egusphere-2024-3619-AC2
-
AC2: 'Reply on RC2', Nicola McDonald, 22 Jan 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 161 | 48 | 8 | 217 | 28 | 10 | 10 |
- HTML: 161
- PDF: 48
- XML: 8
- Total: 217
- Supplement: 28
- BibTeX: 10
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 99 | 46 |
| New Zealand | 2 | 32 | 14 |
| China | 3 | 15 | 7 |
| France | 4 | 8 | 3 |
| Germany | 5 | 6 | 2 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 99