Multi-hazard Tropical Cyclone Risk Assessment for Australia

Do, Cameron; Kuleshov, Yuriy

doi:https://doi.org/10.5194/nhess-2022-139

Preprints

https://doi.org/10.5194/nhess-2022-139

Preprints

19 May 2022

| 19 May 2022

Status: this preprint has been withdrawn by the authors.

Multi-hazard Tropical Cyclone Risk Assessment for Australia

Cameron Do and Yuriy Kuleshov

Abstract. Tropical cyclones (TCs) have long posed a significant threat to Australia's population, infrastructure, and natural environment. This threat may grow under climate change as projections indicate continuing sea level rise and increases in rainfall during TC events. Previous TC risk reduction efforts have focused on the risk from wind alone, whereas a holistic approach requires multi-hazard risk assessments that also consider impacts of other TC-related hazards. This study assessed and mapped TC risk nationwide, focusing on the impacts on population and infrastructure from the TC-related hazards of wind, storm surge, flooding and landslides. Risk maps were created at the Local Government Area (LGA) level for all of Australia, using collated data on multiple hazards, exposure and vulnerability. The study demonstrated that the risk posed by all hazards was highest for coastal LGAs of eastern Queensland and New South Wales followed by medium risk across Northern Territory and north-west of Western Australia, with flood and landslide hazards also affecting several inland LGAs. The resulting maps of risk will provide decision-makers with the information needed to further reduce TC risk, save lives, protect the environment, and reduce economic losses.

This preprint has been withdrawn.

Received: 12 May 2022 – Discussion started: 19 May 2022

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 1407 KB)

Withdrawal notice
This preprint has been withdrawn.
Preprint (1407 KB)

Download & links

This preprint has been withdrawn.

Cameron Do and Yuriy Kuleshov

Interactive discussion

Status: closed

RC1:
'Comment on nhess-2022-139', Anonymous Referee #1, 15 Jul 2022

Dear editor, I have read accurately the manuscript and I found some criticalities that, in my opinion, should be solved before considering it for publication. In the following you can see my comments for the the Authors.

Dear Authors, I have read accurately your manuscript and I found some criticalities that, in my opinion, should be solved.

First of all, at the beginning of the manuscript could be useful a table with the acronyms used in the text.

Section 1

Line 29: please specify the scale of intensity of TCs.

Lines 42-47: I maintain that could be useful a map showing the LGAs and Australian States (WA, NT, SA, QLD, NSW, VIC, and TAS) and some statistics on LGAs (e.g. number of LGAs, minimum, maximum, mean, and median extension). I suggest adding a study area section describing the physical and economic characteristic of Australia and all the toponyms cited in Section 3. Moreover, you do not explain at all the TC risk in Australia. It is important to add a part (may be in the study area section) describing the TC risk in Australia.

Lines 53-68: It is not clear if in your study the analysed risks are caused by TCs. It is a cascade approach or not? I do not understand why and how it is possible to obtain TC risk combining surge, flood, wind, and landslide risks.

Section 2

In my opinion in this section must be added a short part describing the difference among variables, indicators, and indexes. May help citing the “Pyramid of Information” of Hammond et al. (1995) or explaining better the IRDS that you cite in your paper.

Moreover, I maintain that the table in Appendix may be reported in Section 2 and I suggest adding in the table the data format and resolution. In addition, I suggest explaining better these data in Section 2.1.

Lines 70-75: If I understand correctly, you started from hazard, exposure, and vulnerability indicators that were combined to obtain hazard, exposure, and vulnerability indexes by using equal weighting for exposure and Pareto front-ranking for vulnerability (and for risk?). Consequently, I suppose, you obtained surge, flood, wind, and landslide risks. And, finally, combing these risks you obtained TC risk. It is correct? Please explain it better. In my opinion the sentence “This data was then joined to LGA map shapefiles in ArcGIS Pro” may be changed in “This data was then combined to LGA map shapefiles in ArcGIS Pro”, because “join” is a particular GIS command.

I maintain that Figure 1 do not explain in a correct way the risk mapping process. I suggest separating the part of the figure concerning the four term of equation 1 (risk, hazard, exposure, and vulnerability), as well as the different considered risks (surge, flood, wind, landslide, and TC). Moreover, the figure does not explain clearly the process of transforming the indicators to indexes. I suggest modifying the figure and explaining it analytically. Additionally, I suggest explaining briefly the Pareto ranking and the Figure 2 that, in my opinion, it is not clear. Consequently, I suggest re-writing the Section 2.3.

Section 3

The acronyms of the Australian states are not always evident in the maps. Please modify the maps accordingly.

You cited Pibara region, Mount Isa LGA, major coastal cities, major cities, inner cities, mining industries, urban areas, Maralinga Tjarutja LGA, Darwin, Great Diving Range, Tiwi and Mornigton Islnds, Bisbane, Cairns. Where are located these areas?

Overall, in my opinion the manuscript needs to be improved before considering it for publication.

Citation: https://doi.org/10.5194/nhess-2022-139-RC1
- AC1: 'Reply on RC1', Y. Kuleshov, 05 Sep 2022
  
  Dear Reviewer, thank you for your valuable comments which helped us to improve quality of the manuscript. All your comments have been addressed in a revised version of the manuscript . Please find point by point response as well as the revised manuscript attached. We hope you will find this revision satisfactory.
  
  Citation: https://doi.org/10.5194/nhess-2022-139-AC1
RC2:
'Comment on nhess-2022-139', Anonymous Referee #2, 20 Jul 2022
The manuscript provides a framework for comparing TC-related risk across Australia, incorporating multiple hazards, multiple exposure elements and multiple indicators of vulnerability.

The framework described is a relative risk rating, based on a limited view of the probability of events (i.e. a 1% AEP level of hazard) in combination with national-scale indicators of exposure and vulnerability.

Similar efforts have been undertaken within government in recent times, but are as yet unpublished. This manuscript provides a stimulating discussion on the complexity of evaluating multi-hazard risk in a nationally-consistent framework.

Properly undertaken, the resulting information from this analysis could be valuable for prioritising interventions across the country.

The derivation of some metrics warrants further discussion - the range of spatial scales presents unique challenges to developing representative rankings of hazards, especially with relatively coarse information. Flood and storm surge inundation are highly sensitive to spatial resolution, and will be challenging to represent at LGA resolution.

The elements of exposure and vulnerability must be linked - using social vulnerability indicators in combination with physical asset exposure will not produce a valid evaluation of risk (either physical or social).

We have concerns over the evaluation of risk, showing highest risk in northern NSW, including some LGAs well inland where wind hazard will be declining, there is no surge hazard and flood hazard is evaluated to be in the lowest quintile.

Specific comments and questions for the authors:

Line 78: Only one ARI is used - the relative impact may change for different return levels due to the different spatial pattern in hazard and/or vulnerability. Do the authors have any comments on this?

Line 79: It is not appropriate to say this is representative of the hazards in the near future. The 100-year ARI hazard level is an indication of the long term probability in that it occurs - on average - once in every 100 years. There is a possibility of such an event occurring in any given year (approximately a 1% probability), with no inference about the near future likelihood. Further, the 100-year return period level may well change over the next 100 years. Knutson et al (2020) report the most confident TC-related projection is increased storm surge levels, with medium to high confidence that TC-related precipitation will increase at the global scale.

Section 2.1: There is not sufficient discussion on the metrics used for the hazard indices. No references are provided for flood or landslide hazard information in the main part of the manuscript (a table is presented in the appendix, but it is not referenced, and the links in the table are not accessible); the reference provided for Storm Surge does not describe that hazard ("For this global study, the effects are only related to the wind speed at a global scale." Cardona , 2014). This is a major concern to the core objectives of the manuscript.

Line 80: There are more up-to-date sources of information for storm tide hazard - e.g. the Canute 3.0 data available through the NESP Climate Hub (https://shiny.csiro.au/Canute3_0/)

Line 81: Mean values of hazard may not be appropriate for some LGAs. This is an issue the authors note (in reference to East Pilbara). However, the hazard needs to be evaluated in the context of exposed assets. In the case of East Pilbara, the majority of exposed assets (primarily population) are close to the coastline, where wind hazard (and flood hazard) will be higher. In our comparative rankings, we have used a 90th percentile of the hazard level, reflecting the general proximity of population to the coastline.

Line 89: No reference to the table of data sources is provided.

Line 89: Power line and electrical substations will be highly correlated, so using both as input to the exposure definition will be unduly weighted to that infrastructure element.

Line 89: What power line information was used - distribution lines or transmission lines? In some urban LGAs, there may be limited transmission network coverage, with power supplied through lower voltage feeder networks that may lead to biased estimates of exposure. The data table provided does not contain working links, so readers are not able to inspect those sources.

Overall losses will be impacted by the value of lost income to businesses. With no business information included, this may lead to an underestimate of exposure in some areas.

Line 103: The choice of vulnerability indexes is not clearly linked to the choice of exposure indexes. In the Hazard-Exposure-Vulnerability framework, the vulnerability is directly related to the exposed assets. Using social vulnerability indicators and physical assets presents a logical mismatch between the two risk factors. Ideally, physical vulnerability indicators should be used that link the hazard to the exposed physical assets.

The "no vehicle homes" is duplicated in the contributing indicators in the IRSD indicators, so places undue weighting on this indicator. Additionally, the claim of "no vehicle homes" indicator being particularly relevant should be justified - what evidence is there to support the assertion they are more susceptible to loss of life, especially given the very limited fatalities attributable to TCs in Australia? Further, evacuation is only a consideration in storm tide prone areas. Otherwise, the emergency services advice is to shelter in place (i.e. at dwellings that are built to modern codes). A better indicator of vulnerability would therefore be the proportion of houses that are not constructed to modern wind loading standards.

Line 274: The use of data with null values for some LGAs suggests additional effort is required to ensure consistent coverage - either through alternate indexes or suitable estimations from other sources.

Line 307: Correct "main coastland"

Line 377: ABS data would typically be well validated. Engagement with the ABS may have addressed the authors concerns over validation of the (vulnerability) indicators.

Several of the references are incomplete or inaccessible e.g. Scawthorn et al., 2006, Do and Kuleshov, 2022, Burston et al. (missing journal name)

Appendix: None of the links in the table are accessible - appears the links have not been properly included in conversion to PDF. "Geosciences Australia" should be "Geoscience Australia"
Citation: https://doi.org/10.5194/nhess-2022-139-RC2
- AC2: 'Reply on RC2', Y. Kuleshov, 05 Sep 2022
  
  Dear Reviewer, thank you for your valuable comments which helped us to improve quality of the manuscript. All your comments have been addressed in a revised version of the manuscript. Please find point by point response as well as the revised manuscript attached. We hope you will find this revision satisfactory.
  
  Citation: https://doi.org/10.5194/nhess-2022-139-AC2

Interactive discussion

Status: closed

RC1:
'Comment on nhess-2022-139', Anonymous Referee #1, 15 Jul 2022

Dear editor, I have read accurately the manuscript and I found some criticalities that, in my opinion, should be solved before considering it for publication. In the following you can see my comments for the the Authors.

Dear Authors, I have read accurately your manuscript and I found some criticalities that, in my opinion, should be solved.

First of all, at the beginning of the manuscript could be useful a table with the acronyms used in the text.

Section 1

Line 29: please specify the scale of intensity of TCs.

Lines 42-47: I maintain that could be useful a map showing the LGAs and Australian States (WA, NT, SA, QLD, NSW, VIC, and TAS) and some statistics on LGAs (e.g. number of LGAs, minimum, maximum, mean, and median extension). I suggest adding a study area section describing the physical and economic characteristic of Australia and all the toponyms cited in Section 3. Moreover, you do not explain at all the TC risk in Australia. It is important to add a part (may be in the study area section) describing the TC risk in Australia.

Lines 53-68: It is not clear if in your study the analysed risks are caused by TCs. It is a cascade approach or not? I do not understand why and how it is possible to obtain TC risk combining surge, flood, wind, and landslide risks.

Section 2

In my opinion in this section must be added a short part describing the difference among variables, indicators, and indexes. May help citing the “Pyramid of Information” of Hammond et al. (1995) or explaining better the IRDS that you cite in your paper.

Moreover, I maintain that the table in Appendix may be reported in Section 2 and I suggest adding in the table the data format and resolution. In addition, I suggest explaining better these data in Section 2.1.

Lines 70-75: If I understand correctly, you started from hazard, exposure, and vulnerability indicators that were combined to obtain hazard, exposure, and vulnerability indexes by using equal weighting for exposure and Pareto front-ranking for vulnerability (and for risk?). Consequently, I suppose, you obtained surge, flood, wind, and landslide risks. And, finally, combing these risks you obtained TC risk. It is correct? Please explain it better. In my opinion the sentence “This data was then joined to LGA map shapefiles in ArcGIS Pro” may be changed in “This data was then combined to LGA map shapefiles in ArcGIS Pro”, because “join” is a particular GIS command.

I maintain that Figure 1 do not explain in a correct way the risk mapping process. I suggest separating the part of the figure concerning the four term of equation 1 (risk, hazard, exposure, and vulnerability), as well as the different considered risks (surge, flood, wind, landslide, and TC). Moreover, the figure does not explain clearly the process of transforming the indicators to indexes. I suggest modifying the figure and explaining it analytically. Additionally, I suggest explaining briefly the Pareto ranking and the Figure 2 that, in my opinion, it is not clear. Consequently, I suggest re-writing the Section 2.3.

Section 3

The acronyms of the Australian states are not always evident in the maps. Please modify the maps accordingly.

You cited Pibara region, Mount Isa LGA, major coastal cities, major cities, inner cities, mining industries, urban areas, Maralinga Tjarutja LGA, Darwin, Great Diving Range, Tiwi and Mornigton Islnds, Bisbane, Cairns. Where are located these areas?

Overall, in my opinion the manuscript needs to be improved before considering it for publication.

Citation: https://doi.org/10.5194/nhess-2022-139-RC1
- AC1: 'Reply on RC1', Y. Kuleshov, 05 Sep 2022
  
  Dear Reviewer, thank you for your valuable comments which helped us to improve quality of the manuscript. All your comments have been addressed in a revised version of the manuscript . Please find point by point response as well as the revised manuscript attached. We hope you will find this revision satisfactory.
  
  Citation: https://doi.org/10.5194/nhess-2022-139-AC1
RC2:
'Comment on nhess-2022-139', Anonymous Referee #2, 20 Jul 2022
The manuscript provides a framework for comparing TC-related risk across Australia, incorporating multiple hazards, multiple exposure elements and multiple indicators of vulnerability.

The framework described is a relative risk rating, based on a limited view of the probability of events (i.e. a 1% AEP level of hazard) in combination with national-scale indicators of exposure and vulnerability.

Similar efforts have been undertaken within government in recent times, but are as yet unpublished. This manuscript provides a stimulating discussion on the complexity of evaluating multi-hazard risk in a nationally-consistent framework.

Properly undertaken, the resulting information from this analysis could be valuable for prioritising interventions across the country.

The derivation of some metrics warrants further discussion - the range of spatial scales presents unique challenges to developing representative rankings of hazards, especially with relatively coarse information. Flood and storm surge inundation are highly sensitive to spatial resolution, and will be challenging to represent at LGA resolution.

The elements of exposure and vulnerability must be linked - using social vulnerability indicators in combination with physical asset exposure will not produce a valid evaluation of risk (either physical or social).

We have concerns over the evaluation of risk, showing highest risk in northern NSW, including some LGAs well inland where wind hazard will be declining, there is no surge hazard and flood hazard is evaluated to be in the lowest quintile.

Specific comments and questions for the authors:

Line 78: Only one ARI is used - the relative impact may change for different return levels due to the different spatial pattern in hazard and/or vulnerability. Do the authors have any comments on this?

Line 79: It is not appropriate to say this is representative of the hazards in the near future. The 100-year ARI hazard level is an indication of the long term probability in that it occurs - on average - once in every 100 years. There is a possibility of such an event occurring in any given year (approximately a 1% probability), with no inference about the near future likelihood. Further, the 100-year return period level may well change over the next 100 years. Knutson et al (2020) report the most confident TC-related projection is increased storm surge levels, with medium to high confidence that TC-related precipitation will increase at the global scale.

Section 2.1: There is not sufficient discussion on the metrics used for the hazard indices. No references are provided for flood or landslide hazard information in the main part of the manuscript (a table is presented in the appendix, but it is not referenced, and the links in the table are not accessible); the reference provided for Storm Surge does not describe that hazard ("For this global study, the effects are only related to the wind speed at a global scale." Cardona , 2014). This is a major concern to the core objectives of the manuscript.

Line 80: There are more up-to-date sources of information for storm tide hazard - e.g. the Canute 3.0 data available through the NESP Climate Hub (https://shiny.csiro.au/Canute3_0/)

Line 81: Mean values of hazard may not be appropriate for some LGAs. This is an issue the authors note (in reference to East Pilbara). However, the hazard needs to be evaluated in the context of exposed assets. In the case of East Pilbara, the majority of exposed assets (primarily population) are close to the coastline, where wind hazard (and flood hazard) will be higher. In our comparative rankings, we have used a 90th percentile of the hazard level, reflecting the general proximity of population to the coastline.

Line 89: No reference to the table of data sources is provided.

Line 89: Power line and electrical substations will be highly correlated, so using both as input to the exposure definition will be unduly weighted to that infrastructure element.

Line 89: What power line information was used - distribution lines or transmission lines? In some urban LGAs, there may be limited transmission network coverage, with power supplied through lower voltage feeder networks that may lead to biased estimates of exposure. The data table provided does not contain working links, so readers are not able to inspect those sources.

Overall losses will be impacted by the value of lost income to businesses. With no business information included, this may lead to an underestimate of exposure in some areas.

Line 103: The choice of vulnerability indexes is not clearly linked to the choice of exposure indexes. In the Hazard-Exposure-Vulnerability framework, the vulnerability is directly related to the exposed assets. Using social vulnerability indicators and physical assets presents a logical mismatch between the two risk factors. Ideally, physical vulnerability indicators should be used that link the hazard to the exposed physical assets.

The "no vehicle homes" is duplicated in the contributing indicators in the IRSD indicators, so places undue weighting on this indicator. Additionally, the claim of "no vehicle homes" indicator being particularly relevant should be justified - what evidence is there to support the assertion they are more susceptible to loss of life, especially given the very limited fatalities attributable to TCs in Australia? Further, evacuation is only a consideration in storm tide prone areas. Otherwise, the emergency services advice is to shelter in place (i.e. at dwellings that are built to modern codes). A better indicator of vulnerability would therefore be the proportion of houses that are not constructed to modern wind loading standards.

Line 274: The use of data with null values for some LGAs suggests additional effort is required to ensure consistent coverage - either through alternate indexes or suitable estimations from other sources.

Line 307: Correct "main coastland"

Line 377: ABS data would typically be well validated. Engagement with the ABS may have addressed the authors concerns over validation of the (vulnerability) indicators.

Several of the references are incomplete or inaccessible e.g. Scawthorn et al., 2006, Do and Kuleshov, 2022, Burston et al. (missing journal name)

Appendix: None of the links in the table are accessible - appears the links have not been properly included in conversion to PDF. "Geosciences Australia" should be "Geoscience Australia"
Citation: https://doi.org/10.5194/nhess-2022-139-RC2
- AC2: 'Reply on RC2', Y. Kuleshov, 05 Sep 2022
  
  Dear Reviewer, thank you for your valuable comments which helped us to improve quality of the manuscript. All your comments have been addressed in a revised version of the manuscript. Please find point by point response as well as the revised manuscript attached. We hope you will find this revision satisfactory.
  
  Citation: https://doi.org/10.5194/nhess-2022-139-AC2

Cameron Do and Yuriy Kuleshov

Viewed

Total article views: 1,147 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
766	335	46	1,147	37	46

HTML: 766
PDF: 335
XML: 46
Total: 1,147
BibTeX: 37
EndNote: 46

Views and downloads (calculated since 19 May 2022)

Month	HTML	PDF	XML	Total
May 2022	146	44	4	194
Jun 2022	49	6	0	55
Jul 2022	51	21	6	78
Aug 2022	12	19	0	31
Sep 2022	48	24	4	76
Oct 2022	37	10	0	47
Nov 2022	45	7	0	52
Dec 2022	18	9	0	27
Jan 2023	17	8	1	26
Feb 2023	37	13	1	51
Mar 2023	38	20	0	58
Apr 2023	16	37	0	53
May 2023	24	9	0	33
Jun 2023	16	5	0	21
Jul 2023	9	7	2	18
Aug 2023	19	13	1	33
Sep 2023	22	7	0	29
Oct 2023	17	10	1	28
Nov 2023	6	3	0	9
Dec 2023	12	3	1	16
Jan 2024	8	3	0	11
Feb 2024	10	10	5	25
Mar 2024	11	17	0	28
Apr 2024	27	4	4	35
May 2024	13	3	5	21
Jun 2024	7	1	1	9
Jul 2024	11	5	5	21
Aug 2024	16	6	3	25
Sep 2024	12	2	2	16
Oct 2024	7	6	0	13
Nov 2024	5	3	0	8

Cumulative views and downloads (calculated since 19 May 2022)

Month	HTML	PDF	XML	Total
May 2022	146	44	4	194
Jun 2022	49	6	0	55
Jul 2022	51	21	6	78
Aug 2022	12	19	0	31
Sep 2022	48	24	4	76
Oct 2022	37	10	0	47
Nov 2022	45	7	0	52
Dec 2022	18	9	0	27
Jan 2023	17	8	1	26
Feb 2023	37	13	1	51
Mar 2023	38	20	0	58
Apr 2023	16	37	0	53
May 2023	24	9	0	33
Jun 2023	16	5	0	21
Jul 2023	9	7	2	18
Aug 2023	19	13	1	33
Sep 2023	22	7	0	29
Oct 2023	17	10	1	28
Nov 2023	6	3	0	9
Dec 2023	12	3	1	16
Jan 2024	8	3	0	11
Feb 2024	10	10	5	25
Mar 2024	11	17	0	28
Apr 2024	27	4	4	35
May 2024	13	3	5	21
Jun 2024	7	1	1	9
Jul 2024	11	5	5	21
Aug 2024	16	6	3	25
Sep 2024	12	2	2	16
Oct 2024	7	6	0	13
Nov 2024	5	3	0	8

Viewed (geographical distribution)

Total article views: 1,083 (including HTML, PDF, and XML) Thereof 1,083 with geography defined and 0 with unknown origin.

Country	#	Views	%

Cited

Latest update: 17 Nov 2024

Download

This preprint has been withdrawn.

Preprint (1407 KB)
Metadata XML

Short summary

Tropical cyclone (TC) risk over Australia was assessed focusing on the impacts on population and infrastructure from hazards of wind, storm surge, flooding and landslides. The risk posed by all hazards was highest for coastal areas of east, north and north-west of Australia, with flood and landslide hazards also affecting several inland areas. The resulting maps of risk will provide decision-makers with the information needed to further reduce TC risk, save lives, and reduce economic losses.


Total:	0
HTML:	0
PDF:	0
XML:	0

Multi-hazard Tropical Cyclone Risk Assessment for Australia

Interactive discussion

Interactive discussion

Viewed

Viewed (geographical distribution)

Cited

4 citations as recorded by crossref.