1. Introduction

The adverse impacts of landslides on the socio-economic environment are increasing rapidly on a local and global scale. Due to complex physiography and increasing anthropological activities, Himachal Pradesh is recognised for frequent landslide occurrences, especially during the monsoon season. Rainfall induced landslide of Kotrupi on 13 August 2017 was one such devastating incidence. At least 46 people lost their lives as two state transportation busses got buried under a massive landslide along National Highway-154. The highway construction in mountainous regions significantly influences landslide occurrences (Dugonji, 2014). As many highways are under construction and many are in the planning phase in Himachal Pradesh which continuously interfere and destabilise the natural bed slope. The deforestation along the highway alignment also enhances the risk of landslides in Himachal Pradesh.
To quantify and manage the risk due to landslides, the landslide susceptibility analysis and mapping is of utmost importance. Landslide susceptibility of a region can be described as the probability with which a landslide can occur in a region based on that region’s topographical and geographical conditions (Brabb, 1984) and the previous occurrences of landslides in the region (Nohani et al., 2019). Generally, landslide susceptibility is shown on maps that display multi-layered spatial and temporal distribution and the probability with which landslides can occur (Nagarajan et al., 1998). The susceptibility analysis and zonation of landslide hazard is considered as a complex process. It is a readily perceived research area in recent times, and experts have proposed various techniques and methodologies in different geological and meteorological settings (Nayak, 2010; Reichenbach et al., 2018). The multiple stages of landslide risk analysis and management, as compiled by (Fell, 1993), are widely accepted in evaluating and analysing landslide hazard. Various studies have shown the suitability of remote sensing and geomatics-based approaches for susceptibility analysis on a regional scale due to availability of spatially and spectrally varied temporal data (Prakash and Nagarajan., 2018). The remote sensing and GIS-based approaches using high/medium-resolution satellite imagery and Digital Elevation model allows a cost effective and rapid extraction of geological and topographical data on the regional scale (Prakash and Nagarajan., 2017). The first stage in most of the landslide susceptibility mapping methods is to develop a comprehensive landslides inventory map by compiling landslide event data from the various sources for the region. (Pradhan et al., 2014). Landslide hazard assessment and susceptibility analysis are highly dependent on the accuracy of landslides data acquisition and mapping (Henriques et al., 2009; Guzzetti, 2005). Landslide inventory map can be prepared from satellite image by visual interpretation and computer processing of the imagery, field inspections, aerial photographs, high resolution DEM and available data from various published reports (Remondo et al., 2003).
The next phase in landslide susceptibility analysis is to formulate layers of thematic variables of landslide causative factors leading to slope instability. The factors affecting land surface include regional geology, slope curvature, aspect, soil, distance to road, elevation, lineament density, drainage network, NDVI etc. These factors can be extracted and mapped using high resolution remote sensing images, arial photographs, published maps and digital elevation models (DEM) using geospatial tools and techniques. Remote sensing data has been accepted as the most accurate and authentic source of Earth’s surface data (Banshtu and Prakash, 2014). Remote sensing provides the benefit of mapping landside areas according to research demand using updated satellite images (Jaiswal, 2009). These satellite images and aerial photographs being stereoscopic provides three dimensional perspectives for the characterisation of landslides based on their spatial and temporal features of the region (Mantovani et al., 1996; Chakraborthy, 2008). This spatial and temporal thematic dataset needs to be integrated with ground based information (Nagarajan et al., 1998). For this purpose, GIS is a widely accepted tool to store extensive data (Rengers et al., 1992; Soeters et al., 1991). GIS can manipulate and analyse remotely sensed data for assessing landslide hazard (Carrara et al., 1991; McKean et al., 1991). The factors can then be ranked according to available codes, expert opinions, statistical modelling and multicriteria evaluation based techniques (Reichenbach et al., 2018). Hence using GIS in analysing remotely sensed images and DEM’s can be considered as a highly efficient tool for landslide susceptibility mapping (Van Westen, 2000; Jebur et al., 2014b).
Geo-physical laws generally control landslides occurrence, and statistical, empirical and deterministic methods can be used for their analysis (Crozier, 1989; Hutchinson, 1988; Dietrich et al., 1995). The key to predicting landslides of the future is to analyse the past and present scenarios (Varnes, 1984). Along with this, various quantitative and qualitative approaches have been developed, helping determine the frequency and probability of a landslide event (Frangov et al., 2017). Some of the most accurate statistical techniques in landslide susceptibility modelling include logistic regression analysis, data overlay analysis, multi criteria analysis, the weight of overlay analysis, bivariate analysis and entropy based analysis etc. (Huabin et al., 2005; Kanungo et al., 2009; Reichenbach et al., 2018).
The present study is perceived with the motivation that road construction activities significantly influence the study area’s landslide susceptibility. Despite ever increasing landslide incidences along the highways, there is still a significant research gap in comprehensive studies regarding road construction’s impact on landslide susceptibility of a mountainous region like the Mandi District of Himachal Pradesh. Hence this study aims to compare the change in Landslide Susceptibility of Mandi district due to construction of roads in the district using Frequency Ratio (FR), Certainty Factor (CF) and Shanon Entropy (SE) models.