國立虎尾科技大學 |

Identifying Precursors of Creeping Landslides through Remote Sensing and Network Science.

Record Type:	Language materials, manuscript : Monograph/item
Title/Author:	Identifying Precursors of Creeping Landslides through Remote Sensing and Network Science./
Author:	Desai, Vrinda Deepal.
Description:	1 online resource (136 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-06, Section: B.
Contained By:	Dissertations Abstracts International85-06B.
Subject:	Hydrology. -
Online resource:	click for full text (PQDT)
ISBN:	9798381022087

Identifying Precursors of Creeping Landslides through Remote Sensing and Network Science.
Desai, Vrinda Deepal.

Identifying Precursors of Creeping Landslides through Remote Sensing and Network Science. - 1 online resource (136 pages)

Source: Dissertations Abstracts International, Volume: 85-06, Section: B.

Thesis (Ph.D.)--North Carolina State University, 2023.

Includes bibliographical references

As a result of extreme weather conditions, such as heavy precipitation, natural hillslopes can fail dramatically; these can even occur on a dry day due to time lags between rainfall and soil saturation. While the pre-failure deformation is sometimes apparent in retrospect, predicting the sudden transition from gradual deformation to runaway acceleration and catastrophic failure remains challenging. Recent advancements in remote sensing techniques, like satellite radar interferometry, enable high spatial and temporal resolution measurements of deformation and topographic information, providing valuable insights into landslide detection and activity. Landslides are common off the Big Sur coast (Central California) due to active tectonics, mechanically weak rocks, and high seasonal precipitation. Within the study site are 26 active landslides, of which two have slipped - Mud Creek, a slow-moving, deep-seated landslide that catastrophically failed in May 2017 and Paul's Slide, which has experienced lots of shallow slips this year.In this dissertation, I apply methods developed to describe the physics of complex systems to investigate the spatiotemporal patterns of slow deformation for creeping landslides (up to 0.4 m/yr). I transform observations - ground surface displacement (InSAR) and topographic slope (digital elevation model) - into a spatially-embedded network. This data is represented as a multilayer network in which each layer represents a sequential data acquisition period. I use community detection, which identifies stronglycorrelated clusters of nodes, to identify patterns of instability. I have shown that using high-quality data containing information about the rheology (via velocity as a proxy) and susceptibility (slope) of the area successfully forecasts where the Mud Creek failure occurred; this method also identifies two other creeping landslides in the vicinity. Analysis of the overall community structure shows a quiescent period of community persistence that ends in the weeks immediately prior to the failure of Mud Creek.In addition to rheology and susceptibility, I explore how incorporating hydrological information into the multilayer network affects the temporal signal detected by community persistence. I examine different weightings of hydrological information - precipitation from PRISM and soil moisture at different depths and water table depth simulated by WRF-Hydro - to better capture the physical processes leading up to catastrophic failure. The analysis shows that including hydrological information on the types exploreddoes not sufficiently enhance the quality of detected communities or the transition between stable to catastrophic failure. This is because the rheological information already captures the hydrological mechanisms.Having developed these techniques on Mud Creek, whose failure is known, I apply the same techniques on a much larger scale, encompassing 26 active landslides. Using multivariate analysis, I distinguish among a set containing active landslides and control areas using community detection results and information such as precipitation, vegetation, deformation, topography, and satellite coherence. More than 50% of the variation in the dataset is explained using multivariate analysis. I distinguish between landslides that display motion characteristics of approaching catastrophic failure and landslides that are moving but are currently stable. I further classified the active landslides into 4 groups varying in levels of susceptibility and identified a group of landslides that are more prone to unstable behavior, one of which is Paul's Slide, which has since displayed shallow failures. Active landslides that experience high enough deformation that is hard to capture using InSAR move from a moderately susceptible classification to highly susceptible.Through the use of multilayer networks and community detection, I am able to successfully forecast where Mud Creek is, as well as detect the transition from stable to catastrophic failure. The use of rheology and susceptibility are able to encompass the underlying mechanics without added hydrological information. Finally, I apply these techniques to a larger dataset to identify which landslides are high-risk and should be monitored more closely. These observations demonstrate that using network science techniques provides an important advancement in the understanding and forecasting of landslide transitions from slow-moving to sudden acceleration. These observations demonstrate that network science enables us to analyze landslide dynamics in greater depth than simply examining overall deformation.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2024

Mode of access: World Wide Web

ISBN: 9798381022087Subjects--Topical Terms:

642066
Hydrology.
Index Terms--Genre/Form:

554714
Electronic books.

Identifying Precursors of Creeping Landslides through Remote Sensing and Network Science.
LDR:05990ntm a22003497 4500 001 1149667
005 20241022112610.5
006 m o d
007 cr bn ---uuuuu
008 250605s2023 xx obm 000 0 eng d
020 $a 9798381022087
035 $a (MiAaPQ)AAI30727206
035 $a (MiAaPQ)NCState_Univ18402041286
035 $a AAI30727206
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Desai, Vrinda Deepal. $3 1475985
245 1 0 $a Identifying Precursors of Creeping Landslides through Remote Sensing and Network Science.
264 0 $c 2023
300 $a 1 online resource (136 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-06, Section: B.
500 $a Advisor: Leblanc, Sharonda;Wegman, Karl;Riehn, Robert;Daniels, Karen.
502 $a Thesis (Ph.D.)--North Carolina State University, 2023.
504 $a Includes bibliographical references
520 $a As a result of extreme weather conditions, such as heavy precipitation, natural hillslopes can fail dramatically; these can even occur on a dry day due to time lags between rainfall and soil saturation. While the pre-failure deformation is sometimes apparent in retrospect, predicting the sudden transition from gradual deformation to runaway acceleration and catastrophic failure remains challenging. Recent advancements in remote sensing techniques, like satellite radar interferometry, enable high spatial and temporal resolution measurements of deformation and topographic information, providing valuable insights into landslide detection and activity. Landslides are common off the Big Sur coast (Central California) due to active tectonics, mechanically weak rocks, and high seasonal precipitation. Within the study site are 26 active landslides, of which two have slipped - Mud Creek, a slow-moving, deep-seated landslide that catastrophically failed in May 2017 and Paul's Slide, which has experienced lots of shallow slips this year.In this dissertation, I apply methods developed to describe the physics of complex systems to investigate the spatiotemporal patterns of slow deformation for creeping landslides (up to 0.4 m/yr). I transform observations - ground surface displacement (InSAR) and topographic slope (digital elevation model) - into a spatially-embedded network. This data is represented as a multilayer network in which each layer represents a sequential data acquisition period. I use community detection, which identifies stronglycorrelated clusters of nodes, to identify patterns of instability. I have shown that using high-quality data containing information about the rheology (via velocity as a proxy) and susceptibility (slope) of the area successfully forecasts where the Mud Creek failure occurred; this method also identifies two other creeping landslides in the vicinity. Analysis of the overall community structure shows a quiescent period of community persistence that ends in the weeks immediately prior to the failure of Mud Creek.In addition to rheology and susceptibility, I explore how incorporating hydrological information into the multilayer network affects the temporal signal detected by community persistence. I examine different weightings of hydrological information - precipitation from PRISM and soil moisture at different depths and water table depth simulated by WRF-Hydro - to better capture the physical processes leading up to catastrophic failure. The analysis shows that including hydrological information on the types exploreddoes not sufficiently enhance the quality of detected communities or the transition between stable to catastrophic failure. This is because the rheological information already captures the hydrological mechanisms.Having developed these techniques on Mud Creek, whose failure is known, I apply the same techniques on a much larger scale, encompassing 26 active landslides. Using multivariate analysis, I distinguish among a set containing active landslides and control areas using community detection results and information such as precipitation, vegetation, deformation, topography, and satellite coherence. More than 50% of the variation in the dataset is explained using multivariate analysis. I distinguish between landslides that display motion characteristics of approaching catastrophic failure and landslides that are moving but are currently stable. I further classified the active landslides into 4 groups varying in levels of susceptibility and identified a group of landslides that are more prone to unstable behavior, one of which is Paul's Slide, which has since displayed shallow failures. Active landslides that experience high enough deformation that is hard to capture using InSAR move from a moderately susceptible classification to highly susceptible.Through the use of multilayer networks and community detection, I am able to successfully forecast where Mud Creek is, as well as detect the transition from stable to catastrophic failure. The use of rheology and susceptibility are able to encompass the underlying mechanics without added hydrological information. Finally, I apply these techniques to a larger dataset to identify which landslides are high-risk and should be monitored more closely. These observations demonstrate that using network science techniques provides an important advancement in the understanding and forecasting of landslide transitions from slow-moving to sudden acceleration. These observations demonstrate that network science enables us to analyze landslide dynamics in greater depth than simply examining overall deformation.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2024
538 $a Mode of access: World Wide Web
650 4 $a Hydrology. $3 642066
650 4 $a Precipitation. $3 1465176
650 4 $a Failure. $3 1365603
650 4 $a Remote sensing. $3 557272
650 4 $a Landslides & mudslides. $3 1475986
650 4 $a Sand & gravel. $3 1466097
650 4 $a Satellites. $3 683426
650 4 $a Avalanches. $3 1248309
650 4 $a Data assimilation. $3 1475987
650 4 $a Aerospace engineering. $3 686400
650 4 $a Hydrologic sciences. $3 1179176
655 7 $a Electronic books. $2 local $3 554714
690 $a 0799
690 $a 0538
690 $a 0388
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a North Carolina State University. $3 845424
773 0 $t Dissertations Abstracts International $g 85-06B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30727206 $z click for full text (PQDT)