the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 May 2023
Improving fire severity prediction in south-eastern Australia using vegetation specific information
Abstract. Wildfire is a critical ecological disturbance in terrestrial ecosystems. Australia, in particular, has experienced increasingly large and severe wildfires over the past two decades while globally fire risk is expected to increase significantly due to the projected increase in fire weather severity and drought condition. Therefore, understanding and predicting fire severity is critical for evaluating current and future impacts of wildfires on ecosystems. Here, we firstly introduce a vegetation-type specific fire severity classification applied on satellite imagery, which is further used to predict fire severity using antecedent drought conditions, fire weather, and topography of the fire season. Based on a ‘leave-one-out’ cross-validation experiment, we demonstrate high accuracy for both the fire severity classification and the regression using a suite of performance metrics: determination coefficient (R2), mean absolute error (MPE) and root mean square error (RMSE), which are 0.89, 0.05, and 0.07, respectively. Our results also show that the fire severity prediction results using the vegetation-type specific thresholds could better capture the spatial patterns of fire severity, and has the potential to be applicable for seasonal fire severity forecasting due to the availability of seasonal forecasts of the predictor variables.
- Preprint
(8490 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
CC1: 'Comment on nhess-2023-69', Carolina Ojeda, 19 May 2023
This article deals with fire severity prediction in south-eastern Australia using GIS analysis over vegetation indexes. Methodologically, this article is very well written and explained.
My comment is about the main title ¿how this method will improve the fire severity prediction in south-eastern Australia if the authors did not present what the existing systems are?
In the introduction they presented abundant evidence of previous studies done in the region, however, those articles do not address the issue of how in south-eastern Australia the government institutions or private landowners work to predict fire disturbances.
In the discussion, this issue was not addressed and the text was focused on the methodological advantages of this vegetation index over others.
My suggestion is to change the title, adapt it to present those advantages, and leave Australia as a test rather than a true study case.
Citation: https://doi.org/10.5194/nhess-2023-69-CC1 -
CC2: 'Reply on CC1', Kang He, 27 Sep 2023
We appreciate your comments on the paper and suggestions on how to make our study more explicit. Currently, most fire severity prediction models are based on the relationship between fire severity and some strong correlated variables, such as weather, terrain. These studies typically ignore the variability of fire severity from other variables such as landcover, vegetation type and their interaction with climate. Even though there have been some studies focusing on classifying the fire severity level according to vegetation type, these studies have not been connected to fire severity prediction. In our study, we firstly examine how fire severity varies according to vegetation type based on forest fire events in southeastern Australia and then demonstrate through Machine Learning modeling that using the vegetation-specific fire severity classification prior to fire severity prediction could significantly improve the performance of fire severity prediction results. Following the reviewer comments, in the revised manuscript we plan to modify the introduction and discussion sections to better clarify the objectives and unique contributions of this study, and consider modifying our paper title to a more suitable title for the subject matter of this study.
Citation: https://doi.org/10.5194/nhess-2023-69-CC2
-
CC2: 'Reply on CC1', Kang He, 27 Sep 2023
-
RC1: 'Comment on nhess-2023-69', Anonymous Referee #1, 14 Dec 2023
Dear authors, your work focused on the possibility of improving fire severity prediction through specific vegetation information and indexes in a wildfire-affected area in south-eastern Australia. The work is generally well written and I found it interesting.
In any case, different issues need to be considered in your revision:
- I have a first comment about the main focus on fire severity that characterizes your research: fire behavior (that is also described by fire severity), also depends on several other factors that jointly influence it over time. In particular, even if a brief discussion about it is presented in lines 355-363, I suggest better clarifying this issue, especially explaining the relevance of considering all these factors together in fire behavior analysis. For instance, no reference to the importance of the vertical structure of forested areas (DBH, Canopy Cover, CBH, CBD) in this kind of analysis is proposed in the manuscript. Please improve the respective section of the paper by looking at these suggestions.
- Lines 114: here you mention the use of Sentinel 2 together with Landsat 8 data in obtaining pre-NBR. Why did you use both and how you considered the different resolutions of band products in your analysis is not clear or evidenced. Please clarify it by adding an explanation in the methodology section, specifying what satellite data you considered, when, and why also considering the post-processing procedure followed in L8 /S2 data-elaboration. In this regard, you should also improve the Discussion by focusing on other research based on satellite data processing and use in fire-behavior analysis.
- Line 168: why did you choose to consider 20 subsets of fire samples? Please justify this choice.
- Lines 41-54 should be moved to Discussion, where a comparison between your work and other research is needed looking at your paper outline and workflow.
- Please improve the final part of the Discussion citing the possibility to use also different data and tools (such as LiDAR or UAV-based multi-spectral data) in forest fire behavior analysis.
- There are no clear pieces of evidence about future challenges starting from your research. Please enrich the Conclusion in this regard.
Other minor comments are reported below:
- line 17: what did you mean by "fire weather"? please clarify
- line 17: "topography during the fire season". Specify the duration of the fire season and add a reference (what months were considered as fire season?)
- line 22: "forecasting /forecast" repetition. Please change one term
- line 40: add a reference
- Figure 1: increase the size of the legend. Is also not clear if colors are only related to the years or also depends on fire extension (since polygons in the figure are different colored but have also different size). Please specify
- line 95, eq.1: add a reference about dNBR equation
- line 119: is there a repetition of "DEM"? Please clarify since is not clear
- line 124: "wildfire environment": what did you mean with "environment"? Please clarify and rephrase the sentence
- lines 206-213 and line 221: change "figure 2" with "figure 3"
- Figure 3: increase the size of legends
- line 223: add space "were_collected"
- line 231: "Note that" seems quite colloquial, why not change it with something like "is important to consider that" or similar?
- lines 227-231: is not clear how the different percentages were adopted
- Figure 4: legends and descriptions are too small
- Figure 5: as Figure 4
- Figure 6: remove the term "The" in the caption
- Figure 9: legends and items are too small
- lines 338-339: repetition of "method", please rephrase
- line 366: "mis-classification" or "misclassification"?
- line 370: add space: "the_2002"
Good work and best regards
Citation: https://doi.org/10.5194/nhess-2023-69-RC1 - AC1: 'Reply on RC1', Emmanouil Anagnostou, 17 Mar 2024
-
RC2: 'Comment on nhess-2023-69', Anonymous Referee #2, 05 Feb 2024
This paper proposes a novel approach for fire severity, with a focus on the escalating wildfire activity in southern Australia. By introducing a vegetation-type specific fire severity classification method applied to satellite imagery, the paper lays the groundwork for more accurate prediction and assessment of wildfire impacts on ecosystems. The paper is well written and organized, but there are few items that could be addressed to strengthen the importance of the work.
Introduction
The authors state that no classification scheme for southern Australia exists, however literature showed works towards this, see for example (Collins et al., 2018; Dixon et al., 2022; Gale et al., 2023; Gibson et al., 2020). There are also accessible datasets on fire severity available from other sources, for the country, https://datasets.seed.nsw.gov.au/dataset/fire-extent-and-severity-mapping-fesm
Fire severity:
As the technique for dNBR relies on NIR and SWIR, would it be possible to apply the proposed methods to other imagery sources, such as Sentinel or the new Landsat missions? If applicable, it would be beneficial to highlight this point as well for researcher wanting to apply the proposed approach.
Topography:
The authors consider the SRTM as main DEM source, and in the discussion, they highlight how topography appears as an important variable in their model. SRTM however presents limits, especially in areas covered by vegetation, and in general, its error values have strong correlation with terrain slope and certain aspect values (See e.g. (Gorokhovich and Voustianiouk, 2006; Shortridge and Messina, 2011).
For Australia specifically, there is the availability of an upgraded SRTM [SRTM-derived 1 Second -and 3 seconds- Digital Elevation Models Version 1.0, which are an improved DEM compared to the original SRTM. Literature also highlighted that COPDEM30, and the underlying TanDEM-X data, as the most recent and accurate global DEM, and (Hawker et al., 2022) provided a further cleaned version of such a DEM without buildings and Vegetation. Did the authors consider using this upgraded terrain information for the model?
Weather:
How was the ‘1 day window’ decided to get the weather event? Is there a physical meaning linked to this choice or was it operationally decided? I am not sure if it is possible, but have the authors investigated the sensitivity of the results to this window? Literature reported a known potential limitation of the fire history database as the fact that the date of the fire attribute does not always represent the exact burn date (Dixon et al., 2022). Dixon for example proposed a semi-automatic MODIS date-adjustment method to obtain the start and end fire dates: have the authors considered something similar?
Fire severity classes:
As it is my understanding, the severity is based on the dNBR which ranges from -n to +n. Is there a meaningful range of this value representing the severity? (I assume the higher in the positive, the higher the expected impact of the fire -if this is the case, please can you clarify it for the readers not too familiar with the approach? When selecting the quantiles, does the author use the full range of dNBR or focus on a selected part of the distribution (would that matter, if that’s the case?).
I find it a bit confusing that the methods describe a threshold selection, but the whole approach is clarified better in the discussion of the results at chapter 4.2. Would it be possible to restructure a bit this chapter in the method, to clarify how the selection is done?
Maybe this comes from my misinterpretation of the result chapter, but my understanding is that the ground truth for the severity is the ‘observed severity’ from Landsat for some specific fires (Figure 7). If this is the case, and the severity level is defined by a ‘moving’ threshold which in turn is defined by the best model in the training phase, how do you objectively define if the severity is ‘under’ or ‘over’ estimated as compared to the reality of the events? The observed severity is defined using a threshold derived from a ‘training’ of the model.
Would it be possible to compare your severity to some data independent from the threshold choice? I see for example for Australia some other datasets are available, such as
https://data.gov.au/dataset/ds-nsw-c28a6aa8-a7ce-4181-8ed1-fd221dfcefc8/details?q=
Minor comments
Figure 1: it is a bit hard to visualize the ‘wildfire for cross validation’ in the map: is it underlaid to the colored burned areas? I assume the burn years refer to the dataset mentioned in the following page
NSW National Parks and Wildlife Service 88 (NPWS) Fire History – Wildfire and Prescribed Burns dataset (https://data.nsw.gov.au/data/dataset/1f694774-49d5-47b8- 89 8dd0-77ca8376eb04 )
IF so, maybe mention this in the caption.
Also, it appears that the link is not working [I tried and accessed it on 05-feb-2024]
Paragraph from line 206-217: Figure 2 should be Figure 3, Same for the references in the following chapters, it seems the authors refers to figure 3 as 2 (Eg line 221)
Line 212: typo on the number, should be 6.7% not 6,7%
Figure 3: are the vegetation numbers from n to 16 in figure b referring to the legend in figure a? if so maybe leave only one legend to avoid confusion on what the number represents, or add the names of vegetation on the x axis rather than as an additional color bar
References
Collins, L., Griffioen, P., Newell, G., Mellor, A., 2018. The utility of Random Forests for wildfire severity mapping. Remote Sensing of Environment 216, 374–384. https://doi.org/10.1016/j.rse.2018.07.005
Dixon, D.J., Callow, J.N., Duncan, J.M.A., Setterfield, S.A., Pauli, N., 2022. Regional-scale fire severity mapping of Eucalyptus forests with the Landsat archive. Remote Sensing of Environment 270, 112863. https://doi.org/10.1016/j.rse.2021.112863
Gale, M.G., Cary, G.J., van Dijk, A.I.J.M., Yebra, M., 2023. Untangling fuel, weather and management effects on fire severity: Insights from large-sample LiDAR remote sensing analysis of conditions preceding the 2019-20 Australian wildfires. Journal of Environmental Management 348, 119474. https://doi.org/10.1016/j.jenvman.2023.119474
Gibson, R., Danaher, T., Hehir, W., Collins, L., 2020. A remote sensing approach to mapping fire severity in south-eastern Australia using sentinel 2 and random forest. Remote Sensing of Environment 240, 111702. https://doi.org/10.1016/j.rse.2020.111702
Gorokhovich, Y., Voustianiouk, A., 2006. Accuracy assessment of the processed SRTM-based elevation data by CGIAR using field data from USA and Thailand and its relation to the terrain characteristics. Remote Sensing of Environment 104, 409–415. https://doi.org/10.1016/j.rse.2006.05.012
Hawker, L., Uhe, P., Paulo, L., Sosa, J., Savage, J., Sampson, C., Neal, J., 2022. A 30 m global map of elevation with forests and buildings removed. Environ. Res. Lett. 17, 024016. https://doi.org/10.1088/1748-9326/ac4d4f
Shortridge, A., Messina, J., 2011. Spatial structure and landscape associations of SRTM error. Remote Sensing of Environment 115, 1576–1587. https://doi.org/10.1016/j.rse.2011.02.017
Citation: https://doi.org/10.5194/nhess-2023-69-RC2 - AC2: 'Reply on RC2', Emmanouil Anagnostou, 17 Mar 2024
Status: closed
-
CC1: 'Comment on nhess-2023-69', Carolina Ojeda, 19 May 2023
This article deals with fire severity prediction in south-eastern Australia using GIS analysis over vegetation indexes. Methodologically, this article is very well written and explained.
My comment is about the main title ¿how this method will improve the fire severity prediction in south-eastern Australia if the authors did not present what the existing systems are?
In the introduction they presented abundant evidence of previous studies done in the region, however, those articles do not address the issue of how in south-eastern Australia the government institutions or private landowners work to predict fire disturbances.
In the discussion, this issue was not addressed and the text was focused on the methodological advantages of this vegetation index over others.
My suggestion is to change the title, adapt it to present those advantages, and leave Australia as a test rather than a true study case.
Citation: https://doi.org/10.5194/nhess-2023-69-CC1 -
CC2: 'Reply on CC1', Kang He, 27 Sep 2023
We appreciate your comments on the paper and suggestions on how to make our study more explicit. Currently, most fire severity prediction models are based on the relationship between fire severity and some strong correlated variables, such as weather, terrain. These studies typically ignore the variability of fire severity from other variables such as landcover, vegetation type and their interaction with climate. Even though there have been some studies focusing on classifying the fire severity level according to vegetation type, these studies have not been connected to fire severity prediction. In our study, we firstly examine how fire severity varies according to vegetation type based on forest fire events in southeastern Australia and then demonstrate through Machine Learning modeling that using the vegetation-specific fire severity classification prior to fire severity prediction could significantly improve the performance of fire severity prediction results. Following the reviewer comments, in the revised manuscript we plan to modify the introduction and discussion sections to better clarify the objectives and unique contributions of this study, and consider modifying our paper title to a more suitable title for the subject matter of this study.
Citation: https://doi.org/10.5194/nhess-2023-69-CC2
-
CC2: 'Reply on CC1', Kang He, 27 Sep 2023
-
RC1: 'Comment on nhess-2023-69', Anonymous Referee #1, 14 Dec 2023
Dear authors, your work focused on the possibility of improving fire severity prediction through specific vegetation information and indexes in a wildfire-affected area in south-eastern Australia. The work is generally well written and I found it interesting.
In any case, different issues need to be considered in your revision:
- I have a first comment about the main focus on fire severity that characterizes your research: fire behavior (that is also described by fire severity), also depends on several other factors that jointly influence it over time. In particular, even if a brief discussion about it is presented in lines 355-363, I suggest better clarifying this issue, especially explaining the relevance of considering all these factors together in fire behavior analysis. For instance, no reference to the importance of the vertical structure of forested areas (DBH, Canopy Cover, CBH, CBD) in this kind of analysis is proposed in the manuscript. Please improve the respective section of the paper by looking at these suggestions.
- Lines 114: here you mention the use of Sentinel 2 together with Landsat 8 data in obtaining pre-NBR. Why did you use both and how you considered the different resolutions of band products in your analysis is not clear or evidenced. Please clarify it by adding an explanation in the methodology section, specifying what satellite data you considered, when, and why also considering the post-processing procedure followed in L8 /S2 data-elaboration. In this regard, you should also improve the Discussion by focusing on other research based on satellite data processing and use in fire-behavior analysis.
- Line 168: why did you choose to consider 20 subsets of fire samples? Please justify this choice.
- Lines 41-54 should be moved to Discussion, where a comparison between your work and other research is needed looking at your paper outline and workflow.
- Please improve the final part of the Discussion citing the possibility to use also different data and tools (such as LiDAR or UAV-based multi-spectral data) in forest fire behavior analysis.
- There are no clear pieces of evidence about future challenges starting from your research. Please enrich the Conclusion in this regard.
Other minor comments are reported below:
- line 17: what did you mean by "fire weather"? please clarify
- line 17: "topography during the fire season". Specify the duration of the fire season and add a reference (what months were considered as fire season?)
- line 22: "forecasting /forecast" repetition. Please change one term
- line 40: add a reference
- Figure 1: increase the size of the legend. Is also not clear if colors are only related to the years or also depends on fire extension (since polygons in the figure are different colored but have also different size). Please specify
- line 95, eq.1: add a reference about dNBR equation
- line 119: is there a repetition of "DEM"? Please clarify since is not clear
- line 124: "wildfire environment": what did you mean with "environment"? Please clarify and rephrase the sentence
- lines 206-213 and line 221: change "figure 2" with "figure 3"
- Figure 3: increase the size of legends
- line 223: add space "were_collected"
- line 231: "Note that" seems quite colloquial, why not change it with something like "is important to consider that" or similar?
- lines 227-231: is not clear how the different percentages were adopted
- Figure 4: legends and descriptions are too small
- Figure 5: as Figure 4
- Figure 6: remove the term "The" in the caption
- Figure 9: legends and items are too small
- lines 338-339: repetition of "method", please rephrase
- line 366: "mis-classification" or "misclassification"?
- line 370: add space: "the_2002"
Good work and best regards
Citation: https://doi.org/10.5194/nhess-2023-69-RC1 - AC1: 'Reply on RC1', Emmanouil Anagnostou, 17 Mar 2024
-
RC2: 'Comment on nhess-2023-69', Anonymous Referee #2, 05 Feb 2024
This paper proposes a novel approach for fire severity, with a focus on the escalating wildfire activity in southern Australia. By introducing a vegetation-type specific fire severity classification method applied to satellite imagery, the paper lays the groundwork for more accurate prediction and assessment of wildfire impacts on ecosystems. The paper is well written and organized, but there are few items that could be addressed to strengthen the importance of the work.
Introduction
The authors state that no classification scheme for southern Australia exists, however literature showed works towards this, see for example (Collins et al., 2018; Dixon et al., 2022; Gale et al., 2023; Gibson et al., 2020). There are also accessible datasets on fire severity available from other sources, for the country, https://datasets.seed.nsw.gov.au/dataset/fire-extent-and-severity-mapping-fesm
Fire severity:
As the technique for dNBR relies on NIR and SWIR, would it be possible to apply the proposed methods to other imagery sources, such as Sentinel or the new Landsat missions? If applicable, it would be beneficial to highlight this point as well for researcher wanting to apply the proposed approach.
Topography:
The authors consider the SRTM as main DEM source, and in the discussion, they highlight how topography appears as an important variable in their model. SRTM however presents limits, especially in areas covered by vegetation, and in general, its error values have strong correlation with terrain slope and certain aspect values (See e.g. (Gorokhovich and Voustianiouk, 2006; Shortridge and Messina, 2011).
For Australia specifically, there is the availability of an upgraded SRTM [SRTM-derived 1 Second -and 3 seconds- Digital Elevation Models Version 1.0, which are an improved DEM compared to the original SRTM. Literature also highlighted that COPDEM30, and the underlying TanDEM-X data, as the most recent and accurate global DEM, and (Hawker et al., 2022) provided a further cleaned version of such a DEM without buildings and Vegetation. Did the authors consider using this upgraded terrain information for the model?
Weather:
How was the ‘1 day window’ decided to get the weather event? Is there a physical meaning linked to this choice or was it operationally decided? I am not sure if it is possible, but have the authors investigated the sensitivity of the results to this window? Literature reported a known potential limitation of the fire history database as the fact that the date of the fire attribute does not always represent the exact burn date (Dixon et al., 2022). Dixon for example proposed a semi-automatic MODIS date-adjustment method to obtain the start and end fire dates: have the authors considered something similar?
Fire severity classes:
As it is my understanding, the severity is based on the dNBR which ranges from -n to +n. Is there a meaningful range of this value representing the severity? (I assume the higher in the positive, the higher the expected impact of the fire -if this is the case, please can you clarify it for the readers not too familiar with the approach? When selecting the quantiles, does the author use the full range of dNBR or focus on a selected part of the distribution (would that matter, if that’s the case?).
I find it a bit confusing that the methods describe a threshold selection, but the whole approach is clarified better in the discussion of the results at chapter 4.2. Would it be possible to restructure a bit this chapter in the method, to clarify how the selection is done?
Maybe this comes from my misinterpretation of the result chapter, but my understanding is that the ground truth for the severity is the ‘observed severity’ from Landsat for some specific fires (Figure 7). If this is the case, and the severity level is defined by a ‘moving’ threshold which in turn is defined by the best model in the training phase, how do you objectively define if the severity is ‘under’ or ‘over’ estimated as compared to the reality of the events? The observed severity is defined using a threshold derived from a ‘training’ of the model.
Would it be possible to compare your severity to some data independent from the threshold choice? I see for example for Australia some other datasets are available, such as
https://data.gov.au/dataset/ds-nsw-c28a6aa8-a7ce-4181-8ed1-fd221dfcefc8/details?q=
Minor comments
Figure 1: it is a bit hard to visualize the ‘wildfire for cross validation’ in the map: is it underlaid to the colored burned areas? I assume the burn years refer to the dataset mentioned in the following page
NSW National Parks and Wildlife Service 88 (NPWS) Fire History – Wildfire and Prescribed Burns dataset (https://data.nsw.gov.au/data/dataset/1f694774-49d5-47b8- 89 8dd0-77ca8376eb04 )
IF so, maybe mention this in the caption.
Also, it appears that the link is not working [I tried and accessed it on 05-feb-2024]
Paragraph from line 206-217: Figure 2 should be Figure 3, Same for the references in the following chapters, it seems the authors refers to figure 3 as 2 (Eg line 221)
Line 212: typo on the number, should be 6.7% not 6,7%
Figure 3: are the vegetation numbers from n to 16 in figure b referring to the legend in figure a? if so maybe leave only one legend to avoid confusion on what the number represents, or add the names of vegetation on the x axis rather than as an additional color bar
References
Collins, L., Griffioen, P., Newell, G., Mellor, A., 2018. The utility of Random Forests for wildfire severity mapping. Remote Sensing of Environment 216, 374–384. https://doi.org/10.1016/j.rse.2018.07.005
Dixon, D.J., Callow, J.N., Duncan, J.M.A., Setterfield, S.A., Pauli, N., 2022. Regional-scale fire severity mapping of Eucalyptus forests with the Landsat archive. Remote Sensing of Environment 270, 112863. https://doi.org/10.1016/j.rse.2021.112863
Gale, M.G., Cary, G.J., van Dijk, A.I.J.M., Yebra, M., 2023. Untangling fuel, weather and management effects on fire severity: Insights from large-sample LiDAR remote sensing analysis of conditions preceding the 2019-20 Australian wildfires. Journal of Environmental Management 348, 119474. https://doi.org/10.1016/j.jenvman.2023.119474
Gibson, R., Danaher, T., Hehir, W., Collins, L., 2020. A remote sensing approach to mapping fire severity in south-eastern Australia using sentinel 2 and random forest. Remote Sensing of Environment 240, 111702. https://doi.org/10.1016/j.rse.2020.111702
Gorokhovich, Y., Voustianiouk, A., 2006. Accuracy assessment of the processed SRTM-based elevation data by CGIAR using field data from USA and Thailand and its relation to the terrain characteristics. Remote Sensing of Environment 104, 409–415. https://doi.org/10.1016/j.rse.2006.05.012
Hawker, L., Uhe, P., Paulo, L., Sosa, J., Savage, J., Sampson, C., Neal, J., 2022. A 30 m global map of elevation with forests and buildings removed. Environ. Res. Lett. 17, 024016. https://doi.org/10.1088/1748-9326/ac4d4f
Shortridge, A., Messina, J., 2011. Spatial structure and landscape associations of SRTM error. Remote Sensing of Environment 115, 1576–1587. https://doi.org/10.1016/j.rse.2011.02.017
Citation: https://doi.org/10.5194/nhess-2023-69-RC2 - AC2: 'Reply on RC2', Emmanouil Anagnostou, 17 Mar 2024
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 505 | 273 | 43 | 821 | 32 | 34 |
- HTML: 505
- PDF: 273
- XML: 43
- Total: 821
- BibTeX: 32
- EndNote: 34
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1