Landslides in the Nepal Himalaya: a quantitative assessment of spatiotemporal characteristics, susceptibility, and landscape preconditioning

Jones, Joshua (2021) Landslides in the Nepal Himalaya: a quantitative assessment of spatiotemporal characteristics, susceptibility, and landscape preconditioning. Doctoral thesis, University of East Anglia.

[img]
Preview
PDF
Download (43MB) | Preview

Abstract

Mountainous regions such as the Himalaya are severely affected by landslides. Strategies to manage landslide hazard often rely on statistical landslide susceptibility models that forecast the locations of future landslides. Susceptibility models are typically space and/or time independent. However, recent observations suggest that several processes (i.e., earthquake preconditioning, path dependency) are capable of imparting transient controls on landslide occurrence that invalidate the assumption of time-independence. Consequently, it is vital to improve understanding of processes that influence landsliding through space and time, and to assess how these affect typical landslide susceptibility approaches.

Therefore, this thesis aims to quantify the spatiotemporal characteristics, distributions, and preconditioning of monsoon-triggered landslides in the Nepal Himalaya, and how these factors influence regression-based susceptibility modelling. This aim is achieved by developing a 30-year inventory of ~12,900 monsoon-triggered landslides, which is used to: 1) assess the overall characteristics and distributions of monsoon-triggered landsides; 2) systematically quantify spatiotemporal variations in landslide processes and distributions, and how this influences landslide susceptibility modelling; 3) determine empirical relationships between monsoon-strength and landsliding to determine how earthquake preconditioning and cloud-outburst storms transiently perturb landslide rates in Nepal, and 4) recommend a best-practice framework for modelling landslide susceptibility in regions impacted by spatiotemporally varying landslide processes.

Spatiotemporal variations in landslide occurrence are found to relate to permafrost degradation, path dependency, earthquake-preconditioning, and the occurrences of storms. Such variation significantly compromises the applicability and accuracy of regression-based susceptibility models, with models developed from specific regions or time slices incapable of consistently predicting other landslide data. However, susceptibility models developed using 6–8 years of landslide data offered consistently reliable prediction. Overall, it is recommended that typical space-time independent regression-based susceptibility models are avoided in dynamic mountainous regions unless developed with 6-8 years of multi-temporal landslide data and/or specific knowledge of any spatiotemporally varying landslide processes.

Item Type: Thesis (Doctoral)
Faculty \ School: Faculty of Science > School of Environmental Sciences
Depositing User: Chris White
Date Deposited: 28 Apr 2022 09:46
Last Modified: 28 Apr 2022 09:46
URI: https://ueaeprints.uea.ac.uk/id/eprint/84826
DOI:

Actions (login required)

View Item View Item