@article{fdi:010090273,
  title = {{P}recursory motion and time-of-failure prediction of the {A}choma landslide, {P}eru, from high frequency {P}lanet{S}cope satellites},
  author = {{L}acroix, {P}ascal and {H}uanca, {J}. and {A}ngel, {L}. {A}. and {T}aipe, {E}.},
  editor = {},
  language = {{ENG}},
  abstract = {{L}andslide time-of-failure prediction is crucial in natural hazards, often requiring precise measurements from in situ instruments. {T}his instrumentation is not always possible, and remote-sensing techniques have been questioned for detecting precursors and predicting landslides. {H}ere, based on high frequency acquisitions of the {P}lanet{S}cope satellite constellation, we study the kinematics of a large landslide located in {P}eru that failed in {J}une 2020. {W}e show that the landslide underwent a progressive acceleration in the 3 months before its failure, reaching at most 8 m of total displacement. {T}he high frequency of satellite revisit allows us to apply the popular {F}ukuzono method for landslide time-of-failure prediction, with sufficient confidence for faster moving areas of the landslide. {T}hese results open new opportunities for landslide precursors detection from space, but also show the probable seldom applicability of the optical satellites for landslide time-of-failure prediction. {M}any catastrophic landslides are preceded by specific movements, for example, a constant increase in velocity over time. {T}hese accelerations form the basis for predicting the landslide time-of-failure. {T}he measurement of these movements is usually based on instruments installed on the landslide itself. {H}owever, this instrumentation is not always possible and, in recent years, the acquisition of satellite images at increasingly higher frequencies (e.g., {P}lanet{S}cope satellites with a revisit frequency of less than 1 day) has raised interest in landslide monitoring and prediction. {I}n order to assess the capability of {P}lanet{S}cope satellites for this prediction, we analyze here the pre-failure movements of a large landslide in southern {P}eru, which collapsed in {J}une 2020 and dammed a river. {W}e show that the landslide underwent a progressive acceleration in the 3 months prior to its failure, reaching a total displacement of 8 m. {B}ased on the displacement time-series of the fastest areas of the landslide, we predict the landslide time-of-failure 3 weeks in advance. {D}espite these encouraging results for landslide precursor detection from space, this study also shows the seldom use of optical satellites for landslide time-of-failure prediction. {H}igh frequency optical satellites can detect the acceleration precursory to the {A}choma landslide failure{T}he acceleration follows a {V}oight law, classical of the tertiary creep of materials{T}he time-series of ground displacement may be used to predict the time-of-failure for pixels of total displacement larger than 5 m},
  keywords = {landslide ; prediction ; satellites ; time-series ; pre-failure motion ; {PEROU}},
  booktitle = {},
  journal = {{G}eophysical {R}esearch {L}etters},
  volume = {50},
  numero = {19},
  pages = {e2023{GL}105413 [11 p.]},
  ISSN = {0094-8276},
  year = {2023},
  DOI = {10.1029/2023gl105413},
  URL = {https://www.documentation.ird.fr/hor/fdi:010090273},
}