Use of Entropy Theory in Analysis of Landslide Controlling Factors and Zoning of its Occurrence Hazard in Ahmadabad Basin, Lorestan Province

Landslides are one of the most alarming disasters that can cause severe damages to human lives and properties each year. This natural phenomenon can destroy or damage a variety of engineering and human structures, including residential areas, roads, gas pipelines, water and power lines, forests and pastures, agricultural lands and mines. In addition, the social and environmental impacts of this phenomenon such as adverse social effects and increase in the sediment load of rivers should not be overlooked. Due to its mainly mountainous topography, high tectonic activity and seismicity, diverse climatic, and geological conditions, Iran has mainly natural conditions to create a wide range of landslides. So, along with the many natural benefits that exist, the risks involved should not be ignored. The existence of large Zagros faults, alternation of calcareous hard layers and loose marl shale layers in the large anticlines have created favorable conditions for instability of natural slopes throughout Lorestan province. Therefore, this study aimed to prioritize the factors affecting landslide and its risk zonation by using the entropy method in the Ahmadabad basin, Lorestan province.

In this study, 8 factors including slope, lithology, precipitation, land use, elevation, and distance from springs, roads and streams were prepared and quantified as factors affecting landslides. According to the characteristics of the landslides of Ahmadabad basin and expert opinion, each layer was given a score of 1 to 9 based on its importance in occurrence or intensification of landslides. After the classification of layers, the entropy matrix was created. Calculating the entropy matrix and the total weight of the factors (wj) gives the coefficient of occurrence of landslide risk (Hi). The entropy model is expressed as follows.
Ej=-Ki=1npi, jInpi,j                                                                                                                            (1)
pi,j=ri,ji=1mri,j                                                                                                                                           (2)
where, Ej is the entropy value, pij is the decision matrix, and rji is the weight value of each layer, M is the number of landslides, and K is a constant coefficient.
After providing the decision matrix and obtaining the value of Ej, the value of Vj is obtained from Eq. 4.
k=(In m)-1                                                                                                                                         (3)
Vj=1-Ej                                                                                                                                            (4)
Vj is the degree of uncertainty deviation. The Eq. 5 is used to calculate the final weight of all factors (wj).
Wj=Vjj=1mVj                                                                                                                                           (5)
After that, the landslide hazard zonation in the studied basin is evaluated using Eq. 6.
Hi=i=0nWj*ri,j                                                                                                                                (6)

In this research, a bipolar scale is used to convert quantitative and qualitative values. After providing the entropy matrix and converting these criteria to an integer, by calculating the relationships and converting the quantitative values to quantitative ones, the information content in the matrix was first obtained as (pi, j) and, then, the value of Ej is calculated for each factor. The values of Vj and Wj were also calculated using Eqs 4 and 5, and, finally, the regional model of the landslide hazard rate was obtained based on Eq. 6.
Based on the relationship of the landslide risk coefficient, a landslide hazard zonation map has been prepared for the studied basin. Based on the results and in respect with 14.6%, 28.8%, 28.8%, 20.0% and 7.8% of the area were classified as very low, low, medium, high and very high, respectively. Then, the landslide area in each hazard class was determined and the accuracy of the entropy method was estimated using the additive utility method (AUM). The results showed that the accuracy of the method used in predicting landslide hazard zones is about 86%.

In this study, 8 factors have been investigated for zoning landslide risk. Prioritizing the factors affecting landslide using the entropy method shows that three factors including lithology, land use and slope with 24.1, 16.8 and 13.5% of impact have the most important role in landslide occurrence.
Based on the results, more than 56% of the study area is located in medium to high-risk zones, indicating a significant potential for landslide hazards. These zones are mainly located to the north and west of the region, which can be attributed to the sensitive lithology and slope susceptibility of these areas to landslide events. Also, according to the results, 10 villages are located in high to medium risk zones and most of the roads are located in these risk zones.
Therefore, it is necessary to avoid any constructions and increase safety of the existing structures in these areas. The results of the present study showed that the entropy method is suitable for identifying hazardous areas, so it is recommended to use the obtained results in land use decision making and management and regional planning.