Comparative Analysis of Certainty Factor and Analytic Hierarchy Process for Landslide Susceptibility Zonation in Parts of Solan, Himachal Pradesh, India

The landslide inventory map provides an exact idea of event location and frequency of landslides in a particular area. The landslide inventory map can be made from the data that are already available for any area in the environment of ArcGIS. The data of landslides were downloaded from Bhukosh, and the same were verified on Google Earth. A total of 214 landslides were found according to the data.

The slope is considered as one of the most important elements for the occurrence of landslides (Panchal, Shrivastava 2022). It is the changing rate of elevation in the downward or falling direction. The angle of the slope is directly proportional to the stability of the slope (Guo et al. 2021). The slope map for the study area was prepared using the DEM, having a resolution of 30 m (Fig. 2).

Fig. 2.

Curvature refers to the joining of various planes with the surface (Ramesh, Anbazhagan 2015). It is a very essential and important perimeter in the study of the driving and opposing stresses that are contained within the landslide body (Sonker et al. 2021). The classified curvature map is shown in Figure 2.

Aspect defines the direction of a slope with respect to the north direction (Panchal, Shrivastava 2022). The temperature and moisture content are related with the aspect of a slope (Singh, Kumar 2018). An aspect map was also derived from the DEM and was classified (Fig. 2).

Any change in land use and land cover can bring the area under the threat of landslide (Chen et al. 2016). Several studies in the literature have suggested that the role of LULC in bringing about landslides is strong (Singh, Kumar 2018). Toposheets, LISS-III satellite imagery and Google Earth were used to prepare the LULC map (Fig. 2). It is found that the study area has barren land, forest area, agriculture land and populated flat cover in the maximum extent.

Lithology defines the types of rocks present in the area. The resistance to erosion, weathering and slope stability are also dependent on lithology (Bahrami et al. 2021). The datasets required for the lithology map were downloaded from Bhukosh (GSI) and then clipped in ArcGIS software for the particular area that was identified through Google Earth. The map was then prepared and classified (Fig. 2).

Drainage density is defined as the ratio between stream length and the total basin area. Drainage density plays an important role in determining the slope’s groundwater level (Kavzoglu et al. 2014). It has been documented that there is a catastrophic relationship between the drainage density and the ground water conditions. Drainage density is directly proportional to the susceptibility for landslides. The drainage density map was prepared from the DEM of a resolution of 30 m and classified (Fig. 2).

There were four types of soil found in the study area, namely sandy soil, loamy skeletal soil, fine loamy calcareous soil and fine loamy soil. In the area of Kandaghat, it has been found that the soil of the area still retains several minerals that are naturally present in it, among these being phosphorus, zinc, calcium, copper, iron, potassium, magnesium, manganese, nitrogen and organic carbon. The data used for the preparation of soil map were downloaded from Buhkosh (GSI), analysed and prepared using ArcGIS software (Fig. 2).

For this study, rainfall data were downloaded from the CHRS portal (chrsdata.eng.uci.edu/) considering the annual average of 5 years, from 2015 to 2021. Many studies have suggested that rainfall is considered as one of the main triggering factors, acting as the initial trigger point responsible for setting off the landslide event (Kim et al. 2021). As India receives a good amount of rainfall in the monsoon season, Kandaghat, which is situated in the Western Indian Himalayan region, is easily affected by heavy and prolonged rainfall events, which typically result in rock blockages, loss of life and property etc. Thus, rainfall was analysed based on the rainfall map, which was prepared and classified (Fig. 2).

Geological structures such as faults, fractures and rock cleavages play a major role in building the pore pressure. Due to the intense shearing evident from faults in the Himalayan region, there is a threat of landslide hazards (Leir et al. 2004). The lineament density map has been created from the DEM of a resolution of 30 m with the help of ArcGIS software and presented on the classified map (Fig. 2).

Principally applied in adding-up the value outcomes of complex decisions, AHP is considered one of the key tools that can be used in the assessment of the hazards of landslides’ occurrence; it also finds application in relating various phenomena considered potential causative factors for landslides with their actual occurrence (Singh, Kumar 2018). The study and applications of this method have proved to be highly successful in the decision-making process. This method is framed upon three principles, namely: breaking down the complex problem, judgement based on compression and union of respective importance. In the very first step of the AHP method, the causative factors are first systematised to slow hierarchy. The systematised data are then placed in contrast with the matrix as per their possibility of compression, and then the appropriate weights are assigned to each causative factor so that it becomes feasible to compute the consistency ratio (CR) (Singh, Kumar 2018). The weights were assigned inside a comparison matrix to different factors in between the ranges of 1–9 on the basis of their importance. In this process, each given weight undergoes a comparison vis-à-vis the factors that are responsible for causing landslides. Based on the relative significances of the various factors responsible for bringing about landslides, weights were assigned in the compression matrix, using values that lie between 1 and 9, as shown in Table 1. For the purpose of inverse compression, the values of weights ranged in the form of reciprocals of the dominant causative factors. CR can be calculated by the formulae that were given by Saaty (1990):

where:

where:

– λmax = principle Eigen value,

– n is the order of matrix.

Table 1 shows the link between the landslide causative factors and landslide locations using AHP model. The slope angle >45° has a high CF value of 0.48, followed by 35°–45° (0.23), while the remainder among the classes show comparatively low AHP values. Thus, slope angles >45° have high triggering factors and are indicative of greater vulnerability to landslides. The AHP values for the northern part of Kandaghat area show the higher range, followed by the southeast class. The higher the class values of the curvature, the greater the possibilities of landslides. From the present study, we observe that the AHP value of the concave class is demonstrated to be higher at 0.67, followed by those corresponding to the flat and convex ones. Generally, excessive soil moisture levels are associated with an unstable slope and therefore its stability decreases. With the increase in the drainage density, the possibilities of landslides are higher. From the present study, it is observed that the low class shows an AHP value of 0.56, followed by the medium and high ones. Lithology being an internal factor plays a crucial role in governing landslides. The lithology of the present study shows a number of classes in which greywacke and quartzite show higher AHP values of 0.069 and 0.066, respectively. Soil and landslides are inter-linked with each other as any kind of movement in the soil mass can cause a landslide. From the present study, it is observed that loamy soil and loamy sand show the higher values of 0.52 and 0.25, respectively, which indicates that these types of soils have a higher landslide triggering capacity (Table 1).

It has been found that the steepness of a slope is inversely proportional to its stability. Slopes with a steepness of >45° account for approximately 8.92% of the total study area and have the highest AHP weightage of 0.48, with a 35.64% occurrence of landslides. Slopes with angles of 35°– 45° cover 26.34% of the Kandaghat region, and the AHP weightage of landslides that in aggregate occur in such regions is 39.53%. This is followed by slopes having angles ranging from 25° to 35°, which cover 30.11% of the study area and have a weightage of 0.14, with a 20.5% chance of landslide occurrence. If the angle of the slope is <25°, there are fewer instances of landslides. The research area contains slopes with an inclination angle of 35° in 40.10% of the total area, with an 8.94% chance of landslide occurrence. The aspect of a slope refers to the direction it faces in relation to magnetic north. It has been revealed that aspects with a north, northwest, west and southwest orientation have a greater effect on the occurrence of landslides, indicated by the AHP weightage of 0.40, 0.08, 0.14 and 0.23, respectively. Landslides are more likely to occur as a result of water action on surfaces with concave curvature, followed by surfaces with convex curvature. Along the flat curve, the possibility of landslides is eliminated or reduced. The frequency of landslides increased in direct proportion to the density of drainage. Landslides have higher association with areas having more drainage density. Landslides occurring in regions characterised by a high drainage density account for 24.2% of the aggregate landslides occurring in the total study area for the time period under consideration; but despite this fact, areas characterised by a high drainage density amount to only 4% of the total land area. In areas with medium drainage densities, which covered 36.02% of the land, 59.07% of the total landslides occurred. In the 23.62% of the total area characterised by low drainage density, only 25.11% of the total landslides occurred. The density of the lineament had a huge impact on the number of landslides that occurred. The upper class accounts for only 18.82% of the land area, but it is responsible for 59.93% of all landslides. The soil in the present study area has been classified into four categories, namely coarse loamy soil, fine loamy soil, loamy soil and loamy sandy soil; out of these four categories, loamy sandy soil is present mostly on steep slopes, covering an area of 89.92% and being responsible for approximately 93.8% of landslides. On the other hand, fine loamy, loamy soil and coarse loamy soils cover a lesser area and thus their contribution in the occurrence of landslides is lesser. A rainfall of 437–611 mm per year fell on 29.54% of the studied region, resulting in landslides in 27.14% of the area subject to this rainfall. Landslides affected 20.25% of the area with rainfall in the range of 234–437 mm per year. Approximately 23.44% of the area with a rainfall of 745–854 mm per year experienced 57.93% of the landslides. Thus, the higher the rainfall, more likely it is that a landslide will occur. On the other hand, the rest of the classes have comparatively less rainfall and have few chances of landslides. High and medium forests covered approximately 45.2% of the study area and were found responsible for approximately 60.23% of landslide events. Landslides affected 4.97% of barren and agricultural land. Landslides affected 49.8% of the lightly forested and populated area, accounting for 24.35% of the total study area (Fig. 3).

Fig. 3.

This is an indicator used to express or determine the CF (hazard). It is an approach that can be used for the reconciliation of the differences and uncertainties observed in the given (input) values, and also for resolving any issues that might arise pursuant to adding up all the different layers of data (Devkota et al. 2013). The formula for the calculation of CF is defined below:

where Ppa and Pps are the ratio between cell count and landslide count, and ratio of the total cell count to the total landslide count, respectively.

Table 2 defines the calculation of CF for the study area. The CF ranges between −1 and +1, in which the positive values show high certainty and the negative values show low certainty (i.e. an indication that the geological environment is characterised by good conditions). And if the value reaches 0, it then shows that the CF cannot be defined or determined. Table 2 shows the link between the landslide causative factors and landslide location using the CF model. The slope angle >45° has a high CF value of 0.38, followed by 35°–45°, while the remainder of the classes show comparatively low CF values. Thus, the slope angle >45° is responsible for the largest number of potential causal factors, and accordingly, when this angle is prevalent, the vulnerability to landslides is the greatest. The CF values for the southern side show the higher range, followed by the east and western east classes. The higher the class values of the curvature, the greater the possibilities of landslides. From the present study, we observe that the CF value of the convex class is demonstrated to be higher at 0.20, followed by the flat and concave curvature classes. Generally, excessive soil moisture levels are associated with an unstable slope of the curvature zones; the explanation for the association is that when the moisture content of the soil increases, this results in a decrease in its stability. With the increase in the drainage density, the possibilities of landslides are higher. From the present study, we observe that the high class shows a CF value of 0.37, which indicates that there is a high risk of landslides in this class of drainage density. Lithology being an internal factor plays a crucial role in governing landslide activity. The lithology of the present study shows a number of classes in which shale and white-grey quartzite demonstrate the higher CF values of 0.97 and 0.52, respectively.

Soil and landslides are also inter-linked with each other as any kind of movement in the soil mass can cause a landslide. From the present study, it is observed that coarse loamy soil and loamy sand soil show the higher values of 0.9 and 0.5, respectively, which indicates that these types of soils are characterised by a higher landslide triggering capacity (Table 2).

A total of nine landslide causal factors were analysed using the CF method, as shown in Table 2. Slope angle >45° has the highest CF value (0.23), followed by classes of 35–45° (0.32) and 25–35° (0.22), while the remainder of the classes show comparatively low CF, indicating that slope classes having low values are less susceptible and slope classes with higher values are more exposed to the landslide event in the region of Kandaghat. In the case of aspect, south (0.55) shows the highest CF value followed by east and northeast with the CF values of 0.41 and 0.20, respectively. Additionally, it is demonstrated that north, southwest, flat, west, northwest and southeast are characterised by low CF values, which indicates that these classes are less susceptible to landslides. In the case of curvature, the category of convex has the maximum CF value of 0.25; and since the concave and flat classes exhibit negative values, this may be taken as an indication that the occurrence of landslides in these particular classes is low, thereby rendering the concerned regions less prone to landslides. Three classes of lineament density, namely high, medium and low, were considered. The high class has a positive CF value (0.27) and thus is more prone to landslides; and medium and lower classes having the CF values of 0.10 and 0.08, respectively, suggests that there are enough chances of landslides in the medium class as well, but that there is a low possibility of landslides in the low class. With the increase in the drainage density, the possibilities of landslides are higher. Based on scrutiny of the CF for the three classes of drainage density, it is clear that the higher class is having the CF value of 0.34, which demonstrates a significant chance for the occurrence of landslides concerning this class; and as for the other two classes, namely medium and low, having the CF values of 0.10 and –0.1, respectively, it may be inferred that the possibility of landslides is considerably lesser. As mentioned previously, four classified classes of soil are found in the study area; among these, coarse loamy soil has a CF value of 0.92, which indicates that it is imbibed with the highest shearing among these classes, in turn implying the attendant high possibility of landslides; on the other hand, skeletal soil, fine loamy calcareous soil and fine loamy soil have the CF values of 0.52, 0.12 and 0.31, respectively, which suggests a lesser possibility of landslides. The lithology map shows that white quartzite having the high CF value of 0.97 indicates a high possibility of triggering factors for the occurrence of landslide, while white-grey quartzite having the CF value of 0.52, as well as pinkish-grey dolomite shale, cherty dolomite, quartzite, limestone oxidised silt, micaceous sandstone, mudstone, brown sandstone, red clay, purple chocolate shale, shale, slate, quartzite, cherty dolomite clay and purple sandstone having comparatively low values of CF, indicates an absence of landslide.

The rainfall class of 745–854 mm is having the highest CF value at 0.93, which suggests the maximal possibility for landslides; on the other hand, there are comparatively lesser chances for landslides in the range of 611–745 mm that has the CF value of 0.81; further, in the remainder of the ranges, the possibility of landslides’ occurrence is even more sparse. The prepared LULC map is classified into six classes, namely forest area, populated land, less vegetated area, moderately vegetated area, barren land and water bodies; the CF value of 0.80 is the highest sheared by the water bodies, which suggests that the areas near water bodies are highly prone to landslides while other classes are less prone to landslides (Fig. 4).

Fig. 4.