The importance of rainfall infiltration on landslide occurrence at regional scale. Analysis of typhoons in the Philippines
Most landslides occur during or after rainy periods around the world, and many of these have been linked to catastrophic events that resulted in significant property damage and fatalities. In this research project, a physically based model called “Fast Shallow Landslide Assessment Model” (FSLAM) was...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2022 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/375381 |
| Acceso en línea: | https://hdl.handle.net/2117/375381 |
| Access Level: | acceso abierto |
| Palabra clave: | Landslides Typhoons landslide susceptibility Philippines MORLE no-MORLE FSLAM hydrological modelling subsurface hydrology Esllavissades Tifons Àrees temàtiques de la UPC::Enginyeria civil::Geotècnia::Mecànica de roques |
| Sumario: | Most landslides occur during or after rainy periods around the world, and many of these have been linked to catastrophic events that resulted in significant property damage and fatalities. In this research project, a physically based model called “Fast Shallow Landslide Assessment Model” (FSLAM) was used, with a high-resolution topography (5 meters), to map landslide susceptibility for a case study area located in Luzon Island, province of Benguet, Philippines. The research was focused on Typhoon Mangkhut which caused a multiple-occurrence regional landslide events (MORLE) in the area in September 2018. A landslide inventory was collected for this event which was used to assess the performance of the model. Additionally, two no MORLE were tested in July and August 2018. No MORLE were events with higher rainfall intensity than typhoon Mangkhut that did not lead to landslides in the study area. For calibration purpose, an automatic calibration tool (R-FSLAM) was developed in R, which allowed to speed up the calibration process using a multiobjective criteria based on the landslide susceptibility map accuracy of FSLAM. Finally, FSLAM results were coupled with a hydrological model. In terms of statistical performance, FSLAM showed an accuracy of 0.63 during MORLE in September 2018, where stable cells (no-landslide) were better represented (TNR = 0.73) than the unstable cells (no-landslide) (TPR = 0.54). No-MORLE in July and August 2018 performed well reaching an accuracy above 0.9. Two main parameters were found to control cells instability (landslide prone cells) in FSLAM: antecedent effective recharge (q_a) and porosity (n). During MORLE, q_a had to be very low (~ 0.12 mm/day) to cause landslides, while n had to be close to zero. In no-MORLE, q_a was zero and n was greater than n MORLE values. Furthermore, physical meaning of n had to be re-interpretated and renamed (fillable porosity - n_f) as it behaved as a ‘dynamic’ parameter which varies according to soil water |
|---|