prediction of snow avalanche -susceptibility in meigun-shemshak road using rare events logistic regression

1. introduction mountain regions are increasingly endangered by a variety of disaster, including avalanche, debris flows, and rock slide. some characteristics such as elevation, angle of slope, vegetation and climatic elements could affect the avalanche event. indeed, avalanche is the second factor in soil erosion that causes demolition; therefore, it threatens the food safety. iran has two main large mountain ranges of alborz and zagros with risk of avalanches. in mountains and snow routes, avalanche is an inevitable event and due to special geomorphological condition, meigoun- shemshak road is significantly highlighted. meigoun- shemshak road has been situated in shemshak drainage basin. the objectives of this study are to identify the role of effective factors in avalanche events and its zoning in meigoun- shemshak road and to find high risk regions. 2. material and methods meigoun- shemshak road has been located in shemshak drainage basin (51º26'49˝ to 51º31'39˝ e; and 35º57'28˝ to 36º3'15˝ n) with the area of 37.75 km2 in roodbar ghasran district, shemiranat county, northeastern tehran. the minimum and maximum elevations of the study area are 2200 and 4200 m, respectively. figure 1. geographical location of shemshak basin in order to prepare the effective factors in controlling the avalanche events, the primary data and maps including topographic maps in 1:25000 scale and geological maps with 1:100000 scale as well as google earth images were used. independent variables including elevation, aspect, slope, curvature, outcrop rock and geological were prepared based on the basic maps. the snow accumulation area were found using google earth along with the researcher’s knowledge of the region, and then converted to polygon layer by gis software. since the area of avalanche zones is so smaller than the total area, the rare event logistic regression was used to find the effective factor in controlling the avalanche events. roc index was used in order to evaluate the validity of the model, accordingly. finally, the susceptibility map of avalanche events was prepared using the factors resulted from relr. figure 2. variables used in logistic modeling 3. results and discussion the relr was applied to find the relationship between the snow accumulation areas as dependent variable and independent variables, including slope, aspect, geology, elevation and other similar elements. the results showed that depth stratum, curvature, elevation, outcrop rock, slope, and aspect were selected as final factro in controlling the avalanch event. the susceptibility map of avalanche events resulted from controlling factors were classified into three classes, including low risk (67.35% of total area), medium risk (24.9% of total area), and high risk (8.06% of total area). since the area of avalanche regions is less than total area; therefore, we considered 10 % of one and zero regions for the validity model and executed it in half of the basin. the area under the curve is 0.75, and it is reliable for being close to one. table 1. rare event logistic regression results parameter coefficient ᵦ fixed coefficient -6.247 curvature -6.936 elevation 4.436 outcrop rock 1.521 slope 1.254 aspect 1.079 figure 3. regional distribution map affected by avalanche 5. conclusion according to the results of the rare events logistic model, curvature, elevation, outcrop rock, slope and aspect layers were known as effective factors in the snow accumulation area, respectively. 61.49 % of the snow accumulation area is related to convex footslope and 38.51 of density in the concave area. the highest snow accumulation area ranges from 3200 to 3400 elevations and 13% of the basin total area is appropriate for snow accumulation. the slope map indicates that 58.22% of density is in the slope 30 -40º, which is the best slope for the occurance of avalanche. the most snow accumulation area is located in the southern part of the region (49.85% of the total area) and the western part with 29.14 %. 99.97 % of the region, which have no outcrop rocks prone to the snow accumulation area.