Avalanche susceptibility is being mapped out with necessitates taking both surface and climatic conditions into account. Even landscape variables are examined in the study due to their easy accessibility, predictability, and high significance of avalanches. The STRM DEM and Landsat 8 imagery are used In the Shigar valley region, to analyze a variety of input landscape characteristics for avalanche susceptibility mappings, including slope, aspect, curvature, elevation, and terrain roughness, and ground cover. A map of potential avalanche areas is also developed as part of the avalanche inventory. The slope, which is divided into five categories: 12°, 120°–260°, 260°–380°, 380°–510°, and > 510°, is the most important and primary avalanche hazard indicator. Flat, North, East, South, and West are the five types of aspects. Based on observed avalanche sites and ground experience and knowledge, the southern and western slope categories are regarded as more essential for avalanche emergence, and therefore highest points are allotted to these classifications (Table 3).
Concave, flat, and convex are the three types of curvature properties. Convex curvature values have a better chance of avalanche emergence than flat and concave curvature values, according to existing avalanche findings. As a consequence, in the study, convex curvature surpassed flat and concave curvature. As a result, convex curvature outperformed flat and concave curvature in the study (Table 4). The area of investigation terrain is broken down into six categories: 2100, 3200, 4000, 4800, 5600, and > 5600. Maximum avalanches have been observed in height classes of 4800–5600 and < 5600, as per reported avalanche spots, (Sharma and Ganju, 2000). As a result, these categories turn out better than others (Table 3).
The TRI is the research region that has been divided into four categories: low, medium, and high. Following the correlation of terrain roughness levels to known avalanche hotspots and an avalanche prediction, independent professional understanding, the mid and low roughness classes were found to become more essential in avalanche emergence than some other classes of this component. As an outcome, they are rated better than some other roughness classes (Table 3).
The five classes of ground cover maps derived from Sentinel Glacial, unpopulated, residential, water, and grassland/forest/vegetation are all included in the 2A satellite photos. Avalanche formation is aided by snow/ice, barren-land, urban-land, water, and grassland/forest/vegetation cover zones, which are all subject to snowpack uncertainty. As a result, the snow/ice class is ranked higher than most of the others (Table 4). For each eruption event that occurs component, these categorized components MCDA–AHP concept is then applied to stages, given parameter estimate calculated using a density-based clustering comparable array. Table 5 depicts the priority matrix. When compared to other factors in the pairwise choice matrix, the slope component received the highest significance.
The recommendation for each component is expressed as a percentage and is predicated on a 1 to 6 scale (Saaty 1980). Table 2 illustrates this. Snow/ice, barren-land, urban-land, water, and grassland/forest/vegetation cover areas, which are all particularly susceptible to snowpack instability, enable avalanche formation. As an outcome, the snow/ice class is rated higher than the other classes (Table 4). These categorized set of criteria stages are then applied to the MCDA–AHP concept, with a scale factor calculated using a correlation comparison matrix for each accident event. Table 5 shows the optimal grid. The slope feature is used in the bilateral affinity grid. Obtained so much preference because once particularly in comparison to those other variables. The selection for each element is stated as a percentage and is focused primarily on a 1 to 6 scale (Saaty1980). Table 3 illustrates this together.
The most advantageous value the organization aims to compare is the sum of squared variances among measured data every aspect of refinement and quality is approached flexibly. (Jenks, 1967). Jenks optimization is the name given to the Jenks approach. The ASI divides susceptibility into four classifications: safe, low, moderate, and high, (Fig. 10). The high susceptibility area covers 17 percent (1207.207km2) of the research area, according to avalanche susceptibility zones. The study site's moderate susceptibility zones cover 32% of the total area (2267.862 km2).
The remaining low and safe heaven susceptibility territories cover 24% (1721.331 km2) and 27% (1901.01 km2) of the research region, respectively. The distribution of avalanche susceptibility regions in the research area is shown in Table 4.
The Shigar Valley region's southern and northwestern parts are the most vulnerable to avalanches, according to the ultimate avalanche susceptibility map. The Shigar Valley's high susceptibility zones are mostly concentrated in the center, as well as The Shigar River and its successors, for example. Slight susceptibility zones are frequent upon the edges of high susceptibility zones, and they can be found in the lower course, along with those within the mountains (Fig. 10). Even though the MCDA–AHP method has advantages in terms of complicated decision challenges and geographical data layout, evaluating the forecasting rate/accuracy is just as essential for model validation.
The process by ensuring is used to evaluate the MCDA–AHP-based avalanche susceptibility map's forecasting accuracy. The region under the ROC curve, a common analysis inspection technique, is used in this process. One of the most frequently used methods for evaluating predictive accuracy is the ROC-AUC method (Baeza et al. 2010). The prediction rate curve or ROC curve is created by plotting the cumulative end of identified or determined avalanche sites even against estimated ASI, and the region under the ROC curve values is formed (Fig. 13). In a good final result, the point below the curve will be the same as 1. The region beneath a ROC curve with a slope of 0.5 to one is considered to be an incredible suit. While at a cost that is far less than zero. The number five denotes a haphazard record. (Hanley and McNeil 1983) (Fig. 13).