Characterizing the scale of regional landslide  triggering from storm hydrometeorology

Perkins, Jonathan; Oakley, Nina S.; Collins, Brian D.; Corbett, Skye C.; Burgess, W. Paul

doi:https://doi.org/10.5194/nhess-25-1037-2025

Articles | Volume 25, issue 3

https://doi.org/10.5194/nhess-25-1037-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/nhess-25-1037-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 25, issue 3

Research article

| Highlight paper

|

10 Mar 2025

Research article | Highlight paper |

| 10 Mar 2025

Characterizing the scale of regional landslide triggering from storm hydrometeorology

Jonathan Perkins, Nina S. Oakley, Brian D. Collins, Skye C. Corbett, and W. Paul Burgess

Download

Final revised paper (published on 10 Mar 2025)
Preprint (discussion started on 03 Apr 2024)

Interactive discussion

Status: closed

RC1:
'Review of “Characterizing the scale of regional landslide triggering from storm hydrometeorology” by Perkins et al.,', Odin Marc, 22 Apr 2024

The authors present an analysis of several storms induced landslide events, most relating to atmospheric rivers phenomena in California. They retrieve rainfall from a gridded gauge product (covering >10 years, 6hr resolution, 4km spatial resolution), and a leaky bucket model to constrain regolith moisture and derive a soil moisture anomaly, A*, relative to a 15 yr return event. They show that 15 yr return appear to be the minimal return time for causing extensive landsliding based on 4 well constrained cases and then show and discuss the advantage of using a soil moisture anomaly (rather than simple rainfall anomaly) to understand landsliding triggered by rainfall in California. The work is a nice progression from previous work arguing for the use of anomaly to study landslide event (Rainfall anomaly for Marc et al., 2019, or soil moisture anomaly for Saito and Matusyama 2012, but with a more complex methodology and rather preliminary data). Therefore the authors’ work goes provides first basis for simple, physically meainingful and regional scale indicators that could provide a basis for landslide hazard forecasting during storms. In terms of methodology and presentation, I had reviewed a previous version of this work and a lot of my previous concerns in terms of methodology and clarity have been addressed and this version of the draft appears very clear and well thought to me. I therefore congratulate the authors, as I think the work will be a very good contribution to Esurf !
I provide in the attached document a series of minor comments where I have identified potential improvements.
Sincerely, Odin Marc

Citation: https://doi.org/10.5194/egusphere-2024-873-RC1
- AC1: 'Reply on RC1', Jonathan Perkins, 17 Jul 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-873/egusphere-2024-873-AC1-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-873-AC1
RC2:
'Comment on egusphere-2024-873', Ben Mirus, 14 May 2024

This is a nice contribution to the literature in landslide science focused on spatiotemporal assessment of landslide potential using a simplified rainfall index approach. The concept is novel and the manuscript is well written, but some important details are missing related to the simplifying assumptions and methodology, as well as some critical discussion. These gaps can be addressed by addressing my comments in the attached document. Following these revisions the manuscript should ultimately be published as a final paper in NHESS.

Citation: https://doi.org/10.5194/egusphere-2024-873-RC2
- AC2: 'Reply on RC2', Jonathan Perkins, 18 Jul 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-873/egusphere-2024-873-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-873-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Reconsider after major revisions (further review by editor and referees) (25 Jul 2024) by Olivier Dewitte

AR by Jonathan Perkins on behalf of the Authors (05 Sep 2024) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (11 Oct 2024) by Olivier Dewitte

RR by Ben Mirus (04 Nov 2024)

Suggestions for revision or reasons for rejection

The manuscript is considerably improved and the additional analysis strengthens the proof of concept for this method.
Just a few very minor comments here would, in my view, make an iron-clad case for the method and its limitations, or areas for further testing. I will leave it up to the editor as to whether these warrant further delay in publication, or if inclusion of my comments in the discussion is sufficient record for these points.
L202 – Here and anywhere else. AWI Index is redundant in that I = index in the acronym.
L215 - The seasonal reset of AWI is likely ok in the strongly Mediterranean climate of CA, and potentially further north in PNW. However, AK, Appalachia, other parts of the country/world this may not apply, and particularly in a non-stationary world the role of ET will be pivotal. For example, in Colorado, Hinds et al. (2019) found that snowmelt from the previous year was a relevant predictor for landslide activity.
I think the authors need to be careful to mention this in the discussion that the assumption may not be valid everywhere without further testing.
Figure 3 – the black line in 3d is incomplete. Some are boxes (a, c), and the SFBA is a random shape (3b) but why is 3d an incomplete line?
L423-424 - Regarding the spatially variable Kd, the discussion could really include a bit more about this. You have shown with 1 event in California across a nice range of Kd values, but there are so many factors that could be relevant here that I really think you need to here on in the discussion make a more explicit case for further exploration in Kd rather than assuming your approach of picking a value and moving forward has broad applicability.
Great job!

Ben

Hide

RR by Odin Marc (13 Nov 2024)

Suggestions for revision or reasons for rejection

I thank the authors for the detailed answers they gave to my comments and the one of Ben Mirus.

I like the improvement made to figures, and the addition of the sensitivity test to kd ! Basically beyond tiny typo and suggestion I think the paper is ready to become a stimulating piece in NHESS (sorry for confusing with Esurf on the previous review round).

Best regards
Odin MARC

About a discussion with the other reviewer : The case of the length of record and of the timescale representative of extreme event is discussed at length with Reviewer Mirus.

Potentially adding to this discussion I would recall, as discussed in Marc et al 2019, that the recurrence interval used for normalization is likely to work even if NOT the one representative of geomorphic activity :
"the 10-year-return rainfall is a good proxy for the “extreme” climatology over longer time periods (as suggested by its similarity with the 20-year-return rainfall), required for hillslope coevolution through repeated landsliding (e.g., 100 to 1000 years). Indeed, although some landsliding may occur every few years (Imaizumi et al., 2008; Saito et al., 2014), the last landslide event with a magnitude similar to the one of 2011 in the Kii Peninsula occurred ~120 years ago, in August 1889 (Hirano et al., 1984; Inoue & Doshida, 2012)."
In many landscape the spatial pattern of RI for 10, 20 and 50 may be similar and won’t be discriminant for such empirical studies. Then I am not too sure for now what could be the reason of the apparent trend between AWI storm and AWI 10yr in Fig 2. This may just be sample bias (Conchita was really strong?) and be consistent with the fact that Conchita event had landslide density larger to the ones in the 4 other cases… If you have qualitative insights in this direction they may be acknowledged. If not or in the opposite direction, then there is definitely something to explore in a future study…

L89 : True Bogard and Greco 2018 ws not best placed here, but certainly merit to be in the intro, I would place it line 96 with other ref insisting on the role of previous rainfall not only instantaneous rainfall.

L170 : This sentence appears incomplete.

Fig 2 : Great to see the added sensitivity analysis on Kd !

Fig 3 : Thank you for clarifying the meaning of black box / Black lines in the Map panels of Fig 3 ! But it is still not present in the caption ! Please add something like :
« In (a,b,c) black square represent …. and in (d) tha black line represents the GPS tracks from the post-event landslide field verification survey. »

Fig 10c could be in color too no ?

L636 : Should be Fig 10. Nice addition by the way.

L650-655 : I appreciate the reformulation though it does not fully address my concern. I also read again the lines 126-135, which are interesting but mixing quite a lot of things.
Normalizing by extreme rainfall or by mean rainfall can be extremely different in most landscapes… And some process are indeed driven by mean (eg soil production, weathering) while landsliding will be driven by the extreme (that’s why I think putting together Perucacci 2017 and Marc et al 2019 is a bit misleading… ). In Marc et al 2019, we did show that the normalization by mean rainfall is doing significantly worse than normalisation by R10.

In terms of physical explanation, if mean rainfall is higher and enhance soil production this should enhance landslide failure (ie reduce the absolute amount/intensity of rainfall (or AWI) needed to trigger failure… The opposite of what every one reports so far (Perucacci 2017, Marc 2019 and you) so probably not the best exemple of explanation…
In contrast, more extreme rainfall (ie more landsliding) could lead to a thinner soil thickness at geomorphic equilibrium and thus require stronger absolute rainfall to produce as much failures (consistent with observation). Yet this has not been demonstrated (correlating measured soil thickness and the spatial pattern of local climatology). It is also possible that beyond soil thickness ground strength or ground drainage are in cause (and thus the efficiency of the normalization would suggest one or both term must increase with R10 or AWI 10.

So going back to your discussion : what I dislike is that in the past and revised version I feel the text suggest an opposition between two independent (complimentary?) controls on landsliding : the climatology and the « local ground properties » (strength, thickness, drainage etc)… Whereas for me physically the only reason for the climatology (which is also local as far as possible) to improve prediction is because it controls SOME of the local ground properties…
I think we agree on that based on your response to Reviewer Mirus, but that this is not super easy to delineate from L 126-135 nor L 650-55.
So maybe here or there you may attempt some clarification without being too lengthy, or introduce some discussion towards addressing the question of « what ground properties are controlled by the climatology ? » whether with modelling ( ) or empirical studies : eg comparing the extreme climatology (R10 or AWI 10) to soil thickness, or best-fit Kd (from gauge/soil data as discussed by Reviewer Mirus) in similar slope/litho setting and finding a correlation would be amazing!

L926 : Incorrect reference, you use Marc et al 2017 (not 2016) in the text that should point toward :

Marc, O., Meunier, P., and Hovius, N.: Prediction of the area affected by earthquake-induced landsliding based on seismological parameters, Nat. Hazards Earth Syst. Sci., 17, 1159–1175, https://doi.org/10.5194/nhess-17-1159-2017, 2017.

Hide

ED: Publish subject to technical corrections (14 Nov 2024) by Olivier Dewitte

AR by Jonathan Perkins on behalf of the Authors (12 Dec 2024) Manuscript

Executive editor

This paper presents a method for characterizing regional landslide potential, which the authors suggest as an improved basis for landslide hazard forecasting during storms. It discusses the advantage of using considering the relative soil-saturation rather than a rainfall recurrence interval to understand landsliding triggered by rainfall, focusing on data from California.