Rainfall Threshold Estimation and Landslide Forecasting for Kalimpong, India Using SIGMA Model

Abstract

Rainfall-induced landslides are among the most devastating natural disasters in hilly terrains and the reduction of the related risk has become paramount for public authorities. Between the several possible approaches, one of the most used is the development of early warning systems, so as the population can be rapidly warned, and the loss related to landslide can be reduced. Early warning systems which can forecast such disasters must hence be developed for zones which are susceptible to landslides, and have to be based on reliable scientific bases such as the SIGMA (sistema integrato gestione monitoraggio allerta—integrated system for management, monitoring and alerting) model, which is used in the regional landslide warning system developed for Emilia Romagna in Italy. The model uses statistical distribution of cumulative rainfall values as input and rainfall thresholds are defined as multiples of standard deviation. In this paper, the SIGMA model has been applied to the Kalimpong town in the Darjeeling Himalayas, which is among the regions most affected by landslides. The objectives of the study is twofold: (i) the definition of local rainfall thresholds for landslide occurrences in the Kalimpong region; (ii) testing the applicability of the SIGMA model in a physical setting completely different from one of the areas where it was first conceived and developed. To achieve these purposes, a calibration dataset of daily rainfall and landslides from 2010 to 2015 has been used; the results have then been validated using 2016 and 2017 data, which represent an independent dataset from the calibration one. The validation showed that the model correctly predicted all the reported landslide events in the region. Statistically, the SIGMA model for Kalimpong town is found to have 92% efficiency with a likelihood ratio of 11.28. This performance was deemed satisfactory, thus SIGMA can be integrated with rainfall forecasting and can be used to develop a landslide early warning system.

IoT-Based Geotechnical Monitoring of Unstable Slopes for Landslide Early Warning in the Darjeeling Himalayas

“…The study clearly points out to the significance of considering both long-term and short-term rainfall for developing rainfall thresholds for the region. This accounts for the higher number of false alarms reported for the conventional empirical thresholds defined for the study area [8,49]. In 2018, the rainfall received was the lowest when compared to the 2017 and 2019 monsoons.…”

“…From the summary, it is clear that antecedent rainfall plays a major role in ground displacement rather than daily rainfall values. Hence, the study points out the significance of using a statistical threshold that considers the effect of both short-term and long-term rainfall as the first line of early warning [49]. Such thresholds can also be conceptually modified by incorporating moisture content [8,14,50] and tilt angle values.…”

Forecasting Landslides Using Mobility Functions: A Case Study from Idukki District, India

“…In India, the Himalayan belt [4,[22][23][24] and the Western Ghats [25,26] are highly prone to landslides.…”

Definition of 3D rainfall thresholds to increase operative landslide early warning system performances

“…Segoni et al (2018b) pointed out that before the implementation into operational EWS, rainfall thresholds should be carefully validated and evaluated, but this good practice is not fully consolidated. Indeed, some studies conclude that a complete validation shows that the main drawback of the EWS at hand is that a good hit rate is usually achieved at the cost of a high number of false positives (false alarms) (Rosi et al, 2016(Rosi et al, , 2019Abraham et al, 2020;Gariano et al, 2020).…”

“…However, antecedent rainfall and rainfall intensity have been usually considered mutually exclusive, and one of the two approaches is usually selected depending of the characteristics of the test site and the typology of the studied landslides. Other researchers developed complex decisional algorithms based on long-term rainfall anomalies (Martelloni et al, 2012) and observed that in some particular settings (e.g., Emilia Romagna, Italy, and Indian Himalaya) they could outperform simpler approaches based on rainfall intensity, rainfall duration, or event rainfall (Lagomarsino et al, 2015;Abraham et al, 2020). Another series of works, starting from the rationale that landslides are triggered by the increase of water pore pressure and not by rainfall itself, tried to include soil moisture conditions in the threshold modeling (Terlien, 1998;Ponziani et al, 2012;Valenzuela et al, 2018;Wicki et al, 2020).…”