How size and trigger matter: analyzing rainfall- and earthquake-triggered landslide inventories and their causal relation in the Koshi River basin, central Himalaya

Zhang, Jianqiang; van Westen, Cees J.; Tanyas, Hakan; Mavrouli, Olga; Ge, Yonggang; Bajrachary, Samjwal; Gurung, Deo Raj; Dhital, Megh Raj; Khanal, Narendral Raj

doi:https://doi.org/10.5194/nhess-19-1789-2019

Articles | Volume 19, issue 8

https://doi.org/10.5194/nhess-19-1789-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/nhess-19-1789-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 19, issue 8

Research article

|

15 Aug 2019

Research article |

| 15 Aug 2019

How size and trigger matter: analyzing rainfall- and earthquake-triggered landslide inventories and their causal relation in the Koshi River basin, central Himalaya

Jianqiang Zhang, Cees J. van Westen, Hakan Tanyas, Olga Mavrouli, Yonggang Ge, Samjwal Bajrachary, Deo Raj Gurung, Megh Raj Dhital, and Narendral Raj Khanal

Download

Final revised paper (published on 15 Aug 2019)
Preprint (discussion started on 12 Jun 2018)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

SC1: 'A nice systematic analysis of landslide patterns with recognized different triggers', Gianvito Scaringi, 20 Jun 2018
- AC1: 'Response to Dr. Gianvito Scaringi's comments', Jianqiang Zhang, 26 Sep 2018
RC1: 'Review of "How size and trigger matter" by Zhang et al.', Odin Marc, 09 Jul 2018
- AC2: 'Response to RC1 Dr. Odin Marc', Jianqiang Zhang, 26 Sep 2018
RC2: 'Review of How size and trigger matter', Anonymous Referee #2, 13 Jul 2018
- AC3: 'Response to RC2 on Paper nhess-2018-109', Jianqiang Zhang, 26 Sep 2018

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (further review by editor and referees) (20 Dec 2018) by Paolo Tarolli

AR by Jianqiang Zhang on behalf of the Authors (30 Jan 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (04 Feb 2019) by Paolo Tarolli

RR by Odin Marc (16 Feb 2019)

Suggestions for revision or reasons for rejection

2nd Review on Zhang et al., 2019

I have read for a second time the work by Zhang et al.
Overall, the authors have answered many comments to clarify their methods and approach. They also have improved their introduction and discussion.

Still I think 2 points need further clarification/discussion before publication:

1/ The authors better explain the methods and data used to obtain their inventory of RTL. This is an improvement but the author fail to acknowledge the limit of their inventories due to resolution and revegetation issue. This must be stated, with a reference to the recent manuscript by Marc et al., 2019, where in a fraction of the study area of the authors (the Bhothe Koshi valley) hundreds of landslides were mapped between 2010 and 2014, while the authors catalogue (1992-2015) contain only a few in this zone (Fig 3).

2/ I think the author present a somehow biased discussion on the two results of their susceptibility maps : I agree that RTL and ETL susceptibility maps are driven by different factor. In large part because PGA is very important for ETL, not RTL. But RTL susceptibility map is in the end almost not controlled by monsoon long term precipitation, that should be a little discussed by the authors. As a result this map does not really differentiate, in its prediction; RTL or ETL.
So I think the authors need to state and discuss more rigorously their results: RTL susceptibility does not depends much on monsoon, thus either meteorology is less important than in situ parameters either other meteorological parameters should be used (e.g., monsooon variability, cf Deal et al., 2017). This should be explored in future studies. As a result, the RTL maps give a static susceptibility maps, that does not really discriminate landslide from a trigger or another.

Further details are in the specific comments as a well as a fe minor suggestion for clarity and referencing.

Line by line comments

L72 : Please rephrase : They are mostly mapping landslide after an earthquake… But not Marc et al., 2015, that mapped landslide for several years before earthquakes in Taiwan, Papua New Guinea and to a lesser extent Japan.
Also do not cite Marc et al., 2015 for the Wenchuan earthquake that is no considered in this study.
I think Marc et al., 2019 is very relevant to this part of the introduction as it also focussed on RTL before the earthquake and in zones not affected by the earthquake.

L75-76 : Again you misuse Marc et al., 2015. The post earthquake RTL that are reactivation of coseismic landslides are very limited. There does not seem to be a clear correlation with coseismic pattern. The same is observed in Marc et al., 2019 where only 20-30% of RTL caused in 2015, just after the EQ, are spatially connected to ETL.

L95: I think your two questions can be more or less summarized in one : Because if they have the same control, the susceptibility of ETL and RTL should be the same, and one can be used for predicting the other. Opposite is expected if they are controlled by different factor you will not be able to use on to predict the others.

L121 : Maybe mentionning Marc et al., 2019 here would be a good addition, as it gives the magnitude of annual landsliding in different High Himalayan valleys

L141-145 and Fig 3:
RTL 1992 – 2015 : I think that you should mention here that the RTL landslide dataset are undersampling the actual amount of landslide during the 1992-2015 period:
Because of revegetation is rather rapid on landslide scar and possibly because of resolution. An example of that : Marc et al., 2019, mapped between 2010-2014 ~ 350 landslides in a 25x25km² of the Bhote Koshi valley, part of your study area.
Most, if not all of these landlides are not in you 1992-2015 mapping, while your imagery was just a few years after (so resolution or image quality is also likely at play).
If these zones had sustained landsliding at these rates during the last 20 years thousands of landslides are missing in this valley, and likely as much in the other valleys.

It is not necessary a problem for your study, but it is a bias that should be acknowledged, so that readers do not think it is a comprehensive representation of the landslide rate and location. (E.g., the sentence L148 is inaccurate and needs to be removed: “the different images for the period between 1992 and 2015 we were able to recognize most of the landslides”)

L 211 : Ok but the size distribution also depends on how you define landslide area : For example Malamud 2004, explicitly state they remove all debris flow with long aspect ratios. Marc et al., 2019, found a power-law starting at 1000-2000 m3, when considering only landslide scar area (retrieved applying a correction on landslide runout, cf Marc et al., 2018).
>> So maybe mention these separation are relative and maybe improved by removing the area due to runout in individual and stuying landslide scar area only (Marc 2018, 2019)

L337 and Table 2: I am surprised by the statement the RTL reflects monsoonal control: In Table 2 you find a very small control of monsoon (x10 ~1) much smaller than the other types of control such as elevation (x1=7), slope (x2=6) or curvature (x3=-10). So the pattern of RTL

L366 : Also limited because they are not comprehensive : See earlier comment on resolution and revegetation.

L368 : Yes event trigger is a challenge. Note that a first database of RTL event inventories was published recently in Marc et al., 2018.

L386 : “consensus”

L387-389 : Not only topographic, also mechanical properties (as underlined in Frattini and Crosta 2013 or Stark and Guzzetti 2009). Although the methods are somewhat different you could also mention that Marc et al., 2019 found similar Beta values between ETL and RTL, and also relatively similar to your (2.45-2.55).
The cutoff value is much smaller because a correction to remove runout was applied.
This ask the question whether a landslide with a long runout is a large landslide ?

L390 : “ precipitation in the Monsoon for RTL, and PGA distribution for ETL) have major influence on the distribution of landslides and susceptibility zones “
You cannot state that ! This is true for earthquake but NOT for monsoon, see my comment on Table 2 where the effect of monsoon is very small compared to topography, slope, curvature etc.

This is also demonstrated by Fig 10 : ETL susceptibility separate quite well ETL in high susceptibility and RTL in low susceptibility. In contrast, RTL susceptibility does not really distinguish RTL and ETL. This is consistent with the fact that RTL susceptibility is mainly driven by static factor and thus relates to both trigger types.

L402: Ok ETL map is specific, but your statement about RTL are not correct : You state that the RTL predict modestly the ETL… But it predict the small ETL almost as well as the small and large RTL. So you cannot say that RTL has much specificity.

L414 : “cocoseismic” > coseismic
L415 : No Marc et al., 2015 did not say that post-earthquake RTL was following coseismic landslides, see other comments.

Fig 10 : I think some of the values in % are wrong (relative to curve positions...)

References used and not in the manuscript :

Deal, E., Favre, A. C., & Braun, J. (2017). Rainfall variability in the H imalayan orogen and its relevance to erosion processes. Water Resources Research, 53(5), 4004-4021.

Marc, O., Behling, R., Andermann, C., Turowski, J. M., Illien, L., Roessner, S., and Hovius, N.: Long-term erosion of the Nepal Himalayas by bedrock landsliding: the role of monsoons, earthquakes and giant landslides, Earth Surf. Dynam., 7, 107-128, https://doi.org/10.5194/esurf-7-107-2019, 2019.

Marc, O., Stumpf, A., Malet, J.-P., Gosset, M., Uchida, T., and Chiang, S.-H.: Initial insights from a global database of rainfall-induced landslide inventories: the weak influence of slope and strong influence of total storm rainfall, Earth Surf. Dynam., 6, 903-922, https://doi.org/10.5194/esurf-6-903-2018, 2018.

Stark, C. P. and Guzzetti, F.: Landslide rupture and the probability distribution of mobilized debris volumes, J. Geophys. Res.-Earth, 114, F00A02, https://doi.org/10.1029/2008JF001008, 2009.

Hide

RR by Anonymous Referee #3 (21 Feb 2019)

ED: Reconsider after major revisions (further review by editor and referees) (23 Feb 2019) by Paolo Tarolli

AR by Jianqiang Zhang on behalf of the Authors (14 Apr 2019) Author's response Manuscript

ED: Referee Nomination & Report Request started (17 Apr 2019) by Paolo Tarolli

RR by Odin Marc (26 May 2019)

Suggestions for revision or reasons for rejection

3rd Review on Zhang et al., 2019

The authors have answered my previous comments and I have only a few, very minor comments for the paper to be more consistent. I think the paper can be accepted after these minor revisions.

In the abstract I see two sentences that do not seem to correspond anymore to the revised version of the manuscript.

L23 : “The results show that the frequency -area distributions of rainfall-and earthquake–triggered landslides varied considerably, with the former one having a larger frequency of small landslides “

The current text in the result section (p10, 11) and discussion rather reports similar FAD : similar Beta, similar cutoff, and the small extra frequency of ETL at small size may not be reliable given it is well before the cutoff.
So I think you should rephrase to something like : “the frequency -area distributions of rainfall-and earthquake–triggered landslides have similar cutoff and power-law exponent, although the ETL might have a larger frequency of smaller one.”

L31 while susceptibility maps for different size of earthquake-triggered landslides were similar.
This last part of the last sentence does not seem to match the Figure 8/10 and the discussion and conclusion. I suggest removing it.

L72-78 : I appreciate the effort to clarify the debate and statement of the different studies on post-earthquake landsliding.
However I think you should introduce Marc et al., 2019 in the study summary (i.e., L 73 : “Fewer studies … in Japan and Central Nepal, which focus on … Marc et al., 2015, 2019 “)

Further L 78-79 the sentence can be combined (remove the fact it is now not clear which studies give this value of 20-30% of spatial connection).
So I suggest to rewrite : However, other studies argue that there is not a clear correlation of rainfall-triggered landslides with the co-seismic pattern, as only the 20- 30% of the RTL that occurred just after an earthquake, are spatially related to ETL, suggesting limited re-activation of ETL by RTL (Marc et al., 2015, 2019).
With this edit the next sentence (L80) become redundant and can be deleted.

L417 : “ may be more reasonable”

References : Stark and Guzzeti 2009 is added in the bibliography but not in the text. Maybe add it Line 399

Hide

ED: Publish subject to minor revisions (review by editor) (17 Jun 2019) by Paolo Tarolli

AR by Jianqiang Zhang on behalf of the Authors (25 Jun 2019)

ED: Publish as is (10 Jul 2019) by Paolo Tarolli

ED: Publish subject to technical corrections (18 Jul 2019) by Paolo Tarolli

AR by Jianqiang Zhang on behalf of the Authors (23 Jul 2019) Manuscript

Short summary

The aim of this study is to investigate the differences in the mappable characteristics of earthquake-triggered and rainfall triggered landslides in terms of their frequency–area relationships, spatial distributions and relation with causal factors, as well as to evaluate whether separate susceptibility maps generated for specific landslide size and triggering mechanism are better than a generic landslide susceptibility assessment including all landslide sizes and triggers.