the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Brief Communication: Weak correlation between building damage and loss of life from landslides
Abstract. Mapping exposure to landslides is necessary to mitigate risk and reduce vulnerability. Exposure maps can be constructed from building databases, akin to seismic risk assessments, but there has been little investigation of the predictive relationship between building damage and risk to human life from landslides. Our study investigates this relationship globally and in Nepal (47,213 and 5,664 landslides, respectively). While a correlation exists for nationwide totals (R2=0.75), it is near zero for individual events (R2=0.025). It is important not to use building datasets in isolation for landslide exposure maps and disaster planning to avoid unintentionally prioritising building damage over human lives.
- Preprint
(694 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on nhess-2024-40', Daniel Costantini, 09 Apr 2024
Interesting statistical analyses and results!
Below are a few comments that I have noticed:
- Lines 20-24: I would not say that seismic risk maps are constructed only on the basis of the type and condition of buildings. Rather, the seismic microzonation or the type and condition of the soil plays a central role (cf. various earthquakes such as the one in Emiglia Romagna or Abruzzo). I would suggest supplementing this.Â
- Lines 78-80: That's an interesting observation. However, I believe that this observation also has a lot to do with the type of process. For example, when a large slide is activated, you have often little human loss and a lot of damage to buildings. This is not the case with rockfalls, however, where there is often a high level of human loss and less damage to buildings, because these tend to be more localised. I would suggest that the type of landslide process must also be taken into account in any case.
- Further observation: It would be interesting to carry out such statistical analyses on databases that have a high level of detail in terms of geographical localisation and content/description of the event in order to compare whether the results are similar. One example is the IdroGEO platform in Italy - see, for example, the landslides of the Autonomous Province of Bolzano. We are at your complete disposal for a discussion of the analysis of our detailed datasets.
Citation: https://doi.org/10.5194/nhess-2024-40-RC1 - AC1: 'Reply on RC1', Maximillian Van Wyk de Vries, 20 Nov 2024
-
RC2: 'Comment on nhess-2024-40', Anonymous Referee #2, 19 Aug 2024
This study focuses on the relationship between building damage and human loss associated with landslides. By examining a global database, the authors show there is only a weak relationship between building damage and human loss for individual events. There is a much stronger correlation when data are averaged for individual countries. The lack of a strong correlation could have implications for the generation of exposure maps since these could be defined based on building databases. The paper is well written and easy to follow. I have mostly minor comments on methodology and potential discussion points.
1. Comparing and contrasting landslide and earthquake hazards in this context is interesting. One potential discussion point that is not mentioned is that there are early warning systems in place for some types of landslides in many countries but there are no analogous earthquake early warning systems. Since there are early warning systems for some landslides, could this contribute to a weaker correlation between building damage and human loss due to evacuation and improved awareness of the timing of the hazard? For example, we might expect there to be substantial building damage and minimal human loss in a residential area where there is a successful evacuation before a landslide. An exposure map made prior to this event that was based on a building database could still be a good indicator of loss of life in this case if the map was designed to represent a scenario without an evacuation. The potential for evacuations and increased awareness of the timing of the hazard could be a potential explanation for some of the points in figure 1b where there is substantial building damage and negligible human loss.
2. There appear to be a number of points in figure 1b where there is human loss but negligible (i.e. 0 or 1) building damage. One interpretation is that these cases are associated with events in areas that are not residential, such as tourist locations or camping sites where there are people but no or very few permanent structures. I’m sure there could be a variety of other explanations for this as well, but discussing those possibilities could help place the weak correlation in better context since using a building database for exposure maps would only make sense and be possible in cases where there are buildings. When examining the strength of the correlation between building damage and human loss within the context of implications for exposure maps, it would be ideal to only include events that occurred in areas with permanent structures. I understand that this is probably not feasible with the type of data being used here but some discussion around this and related points could be helpful.
Line 51: Is this a linear regression as mentioned in the results? Why only explore a linear relationship between building damage and human loss? Based on figure 2a, a power law might fit better and could be worth exploring. If not, it could also be interesting to report a spearman correlation coefficient. Either way, could you elaborate on the reasoning for expecting a linear relationship if that is the case.
Line 66: Can you comment on the performance of the linear fit based on the distribution of the residuals? Â
Line 87-90: This gets back to the above comments about potential explanations for high-death, low building damage events (e.g. events that occur in areas without permanent structures where building databases wouldn’t be a reasonable possibility for creating an exposure map) and low-death, high building damage events (e.g. events where an evacuation or early warning minimized human loss). I think this is worth discussing since these types of events seem to have a strong effect on the strength of the correlation.
Line 141-143: The strength of the correlation for other hazards is mentioned in the conclusions but not in the results section. I think it would be appropriate to add a little more about these other correlations to the results section.
Citation: https://doi.org/10.5194/nhess-2024-40-RC2 - AC2: 'Reply on RC2', Maximillian Van Wyk de Vries, 20 Nov 2024
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
454 | 131 | 41 | 626 | 38 | 34 |
- HTML: 454
- PDF: 131
- XML: 41
- Total: 626
- BibTeX: 38
- EndNote: 34
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1