Estimation of Rainfall-Induced Landslides Using the TRIGRS Model

Publication Type:
Journal Article
Citation:
Earth Systems and Environment, 2019, 3 (3), pp. 575 - 584
Issue Date:
2019-12-01
Filename Description Size
Dikshit2019_Article_EstimationOfRainfall-InducedLa.pdfPublished Version3.11 MB
Adobe PDF
Full metadata record
© 2019, King Abdulaziz University and Springer Nature Switzerland AG. Rainfall-induced landslides have become the biggest threat in the Indian Himalayas and their increasing frequency has led to serious calamities. Several models have been built using various rainfall characteristics to determine the minimum rainfall amount for landslide occurrences. The utilisation of such models depends on the quality of available landslide and rainfall data. However, these models do not consider the effect of local soil, geology, hydrology and topography, which varies spatially. This study is to analyse the triggering process for shallow landslides using physical-based models for the Indian Himalayan region. This research focuses on the utilisation and dependability of physical models in the Kalimpong area of Darjeeling Himalayas, India. The approach utilised the transient rainfall infiltration and grid-based regional slope-stability (TRIGRS) model, which is a widely used model in assessing the variations in pore water pressure and determining the change in the factor of safety. TRIGRS uses an infinite slope model to calculate the change in the factor of safety for every pixel. Moreover, TRIGRS is used to compare historical rainfall scenarios with available landslide database. This study selected the rainfall event from 30th June to 1st July 2015 as input for calibration because the amount of rainfall in this period was higher than the monthly average and caused 18 landslides. TRIGRS depicted variations in the factor of safety with duration before, during and after the heavy rainfall event in 2015. This study further analysed the landslide event and evaluated the predictive capability using receiver operating characteristics. The model was able to successfully predict 71.65% of stable pixels after the landslide event, however, the availability of more datasets such as hourly rainfall, accurate time of landslide event would further improve the results. The results from this study could be replicated and used in other unstable Indian Himalayan regions to establish an operational landslide early warning system.
Please use this identifier to cite or link to this item: