Use this URL to cite or link to this record in EThOS:
Title: Statistical patterns in triggered landslide event inventories and their use in a landslide-road impact model
Author: Taylor, Faith Elizabeth
ISNI:       0000 0004 8505 1770
Awarding Body: King's College London
Current Institution: King's College London (University of London)
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis examines landslide inventories, their statistical behaviour, and develops a generally applicable landslide-road impact model exploiting this statistical behaviour. Chapter 2 reviews landslide inventories, illustrating how production methods introduce uncertainty. Chapter 3 applies a Boolean search to an archive of 568 UK regional newspapers to identify landslide information to add to the Great Britain National Landslide Database. Results for 2012 [2006] returned 73 [39] database additions. Chapter 4 reviews generally applicable statistical models of triggered landslide event inventory area (AL). Chapter 5 investigates landslide shape using two large, substantially complete triggered-event inventories: 11,111 earthquake-triggered landslides (1994 Northridge, USA); 9,594 rainfall-triggered landslides (1998, Guatemala) and five additional 'lower confidence' inventories. Landslide polygons are abstracted to ellipses to calculate length-to-width ratios (L/W). Maximum-Likelihood Estimation bootstrapping techniques and Monte Carlo Kolmogorov-Smirnov testing show that an inverse-gamma pdf is a good general statistical model of landslide L/W when separated into categories of AL. Chapter 6 investigates spatial clustering in triggered-event inventories using pair-correlation to measure distances between all landslide centroid pairs and a moving-window technique to investigate linkages between landslide susceptibility and clustering. We find similarities in clustering across eight landslide inventories. Chapter 7 exploits triggered landslide event statistics in a landslide-road impact model. Landslide areas and shapes are randomly selected from general statistical distributions and semi-randomly dropped over three regions (Northridge, USA; Collazone, Italy; Shu-Wa, Taiwan) conditioned by susceptibility. The resultant synthetic triggered-event inventories are overlaid with road maps and landslide road blocks identified. Model scenarios are run 100 times (Monte Carlo simulation) to create probabilistic forecasts, then confronted with observed triggered-events. By adjusting road corridor susceptibility, the model closely reflects observed road blockages numbers. This landslide-road impact model presents a low-data methodology to simulate simultaneous road network impacts.
Supervisor: Malamud, Bruce Douglas ; Baas, Andreas Cornelis Wilhelmus Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available