Determination of the Probabilities of Landslide Events—A Case Study of Bhutan

Raju Sarkar,Kelzang Dorji

doi:10.3390/hydrology6020052

Raju Sarkar, Kelzang Dorji

Open Access

https://doi.org/10.3390/hydrology6020052

Copy DOI

Abstract

Landslides have been and are prominent and devastating natural disasters in Bhutan due to its orography and intense monsoonal rainfall. The damage caused by landslides is huge, causing significant loss of lives, damage to infrastructure and loss of agricultural land. Several methods have been developed to understand the relationship between rainfall and landslide incidences. The most common method to understand the relationship is by defining thresholds using empirical methods which are expressed in either intensity-duration or event rainfall-duration terms. However, such thresholds determine the results in a binary form which may not be useful for landslide cases. Apart from defining thresholds, it is significant to validate the results. The article attempts to address both these issues by adopting a probabilistic approach and validating the results. The region of interest is the Chukha region located along the Phuentsholing-Thimphu Highway, which is a significant trade route between neighbouring countries and the national capital Thimphu. In the present study, probabilities are determined by Bayes’ theorem considering rainfall and landslide data from 2004 to 2014. Singular (rainfall intensity, rainfall duration and event rainfall) along with a combination (rainfall intensity and rainfall duration) of precipitation parameters were considered to determine the probabilities for landslide events. A sensitivity analysis was performed to verify the determined probabilities. The results depict that a combination of rainfall parameters is a better indicator to forecast landslides as compared to single rainfall parameter. Finally, the probabilities are validated using landslide records for 2015 using a threat score. The validation signifies that the probabilities can be used as the first line of action for an operational landslide warning system.

Highlights

Landslides are the most common and one of the prominent natural hazards in steep terrain affecting human lives and property, and often leading to loss of lives
Physical models require the use of several parameters involving slope-stability modelling and acquiring the data required to carry out such an analysis are usually challenging [3]
The rainfall intensity rainfall event is defined as the total number of consecutive days of rainfall

Summary

Introduction

Landslides are the most common and one of the prominent natural hazards in steep terrain affecting human lives and property, and often leading to loss of lives. In [1] the study of a global database of landslide occurrences between 2004–2016 showed that 75% of landslides occurred in Asia, with significant occurrences in the Himalayan arc. The study showed that the majority of the landslides are shallow landslides and are triggered by rainfall. To understand the phenomena of shallow landslides, several models were developed and used in several parts of the world. These models can be broadly categorised as empirical or physical models.

Methods

Results

Conclusion