Identifying Effective Factors and Landslide Risk Zoning using the Maximum Entropy Method(Case study: Chalus Watershed)

Abstract

Landslides are one of the most destructive types of domain movements and instabilities, which always cause soil erosion, produce sediment, destroy agricultural lands, gardens, and roads, and cause significant human and financial losses in different parts of the world. Especially in Iran due to the special conditions of the geological structure.

The watershed of Chalus River is located in the northern slope of Central Alborz and in the south of the city in the geographical longitude of 51°, 00'east to 51°, 35' east and latitude 36°, 08' north to 36°, 43' north. The studied basin of Chalus River watershed leads from the west to the Sardabroud River watershed, from the east to the Korkorsar River watershed, from the south to the Karaj watershed, and from the north to the Mazandaran Sea.

In this research, the most important factors affecting landslides have been investigated, which include topography, climate, geology, soil science, land use, distance from the river, topographic humidity index, and vegetation index.

After determining the most important factors affecting landslides in the studied watershed, a layer map was prepared. Then, by using the maximum entropy algorithm with the help of MaxEnt software, one of the capabilities of this model is to identify the most important influencing variables and determine the relative importance of each of the factors affecting the identification of landslide areas and analyze the sensitivity of the model using the Jackknife method.

In this method, after creating a complete model with the involvement of all variables, the modeling is repeated for the number of variables and each time one of the variables is removed from the modeling process. In this way, the effect of each variable in predicting the desired areas was evaluated. Then, in order to evaluate the model, the ROC curve was used, and the area under the obtained AUC chart was taken into consideration as a criterion of the model's discriminating power in detecting presence and non-presence points. In the next step, in order to prepare the stability index, SINMAP plugin and Arc view software were used. Then, by accepting the default values to recalibrate the parameters and apply the corresponding settings and values, the stability index was extracted. In the last stage of the current research, based on the effective factors in the Arc GIS10.8 software environment, a map of the risk of landslides in the Chalus watershed was prepared.

According to the results of the model, the most effective factors in the occurrence of landslides in the study area were rainfall factors, soil science, geological units, slope percentage, land use and distance from the river.

In the present study, AUC chart was used to validate the model. The number of output diagrams of the model is equal to the number of iterations of the model. Finally, the average of model iterations was considered as ROC diagram to evaluate the validity of the model.

The value of AUC for landslide validation is 0.73, which indicates the acceptable prediction and modeling of landslides by the model in the study area.

According to the results obtained from combining the results of the previous sections, the final map was prepared, which according to the findings of the research, the area and percentage of each of the landslide risk classes in the study area were obtained. In the landslide risk zoning map, the sensitivity of the region to the occurrence of this natural phenomenon was determined between zero and one.

Due to the special geographical situation of Iran, each year the phenomenon of landslides imposes a lot of human and financial losses on our country, and one of the ways to reduce these losses is to identify areas prone to landslides with forecasting and zoning methods and providing implementation solutions. In general, it can be said that in all areas where the risk of landslides is more likely, there are formations with low resistance, suitable slopes to provide a landslide bed, and landslide prone landforms. Due to the fact that the role of each factor depends on other effective factors, so its role in the occurrence or non-occurrence of landslides is not the same, therefore, the combination of factors has created a suitable platform for the occurrence of this natural phenomenon. In this regard, in this study, with the aim of zoning landslide risk using the maximum entropy method in the Chalus watershed, it was planned that according to the results obtained from the current study, the risk classes were low, relatively low, medium, relatively high and high, respectively 13.29 , 18.57, 23.73, 35.90 and 8.49 percent of the studied area, which indicates the high potential of the area to cause landslides, so the results of this research can be useful to managers and planners. It helps a lot so that they can make better decisions based on location data.