نوع مقاله : مقاله پژوهشی
1 دانشجوی دکتری مدیریت حوزههای آبخیز، ارومیه، دانشگاه ارومیه، ارومیه، ایران
2 دانشیار، گروه مرتع و آبخیزداری، دانشکده منابع طبیعی دانشگاه ارومیه، ارومیه، ایران
عنوان مقاله [English]
Mass movements (landslides, creeps, falls, mudflows, etc.) as one of the natural disasters, many environmental and human factors can affect their occurrence. The role of us humans in the occurrence of these movements in relation to environmental factors such as tectonics, rainfall, altitude, slope, etc. is very small. On the contrary, it is very much related to human factors such as land use change, irrigation and road construction, which are among the most important of these factors.
Landslides are generally complex geomorphological features. Landslide sensitivity analysis requires consideration of their environmental context. Given the importance of spatial variables in natural hazard management, a large number of researchers have investigated the effects of these factors in modeling landslide risk and sensitivity. Landslides are one of the natural hazards that most people around the world suffer a lot of human and financial losses due to this phenomenon. Landslide risk management is critical to reduce the number of occurrences due to a set of topographic, geoenvironmental, hydrological and geological conditions.
Ziveh watershed with an area of 21686 hectares is located in West Azarbaijan province, southwest of Urmia city and in Margor Silvana section. This basin has an average height of 2265 meters, a minimum height of 1500 meters, a maximum height of 3479 meters and an average slope of 16.5 degrees. The average annual rainfall is estimated at 395 mm.
Through extensive field visits and GPS, the location of landslides that could be accessed for recording was first captured as a polygon. The rest were recorded through Google Earth images. A total of 167 landslide polygons were identified in the area. By reviewing internal and external sources along with notes on the factors affecting the occurrence of each slip during the harvest of information layer points, a total of sixteen environmental, human and morphometric indicators were considered. Digital elevation model map with a resolution of 12.5 12 12.5 meters from the University of Alaska site. Land use map was prepared from the archives of the General Directorate of Natural Resources of West Azerbaijan Province and modified using Google Earth images. Fault data and petrology of the region from the geological map of Silvana sheet (1: 100000), and from the statistics of 13 stations around the basin of the rainfall map using simple kriging interpolation method considering the common statistical period Prepared in 1392-1368.
Maps of environmental and human factors in ArcGIS10.5 environment and maps of geomorphometric indices in SAGA_GIS.6.4 were prepared through digital elevation model with pixel size (12.5 * 12.5). After obtaining the weight of each factor using Shannon entropy models, WOE and LNRF of landslide susceptibility zoning maps were prepared and evaluated using ROC curve.
Results and Discussion
The results of determining the most important factors affecting the occurrence of landslides and its relationship with effective factors in both WOE and LNRF models showed that Shannon entropy model in relation to identifying the most important factors influencing landslide occurrence and sensitivity zoning Landslides have been very effective in Ziveh watershed. According to the results of Shannon entropy model, the factors of topographic position, distance from waterway, distance from fault and land use had the greatest impact on the occurrence of landslides and the least effect related to annual rainfall, profile curvature and altitude. Investigating the weight of subclasses based on the frequency of the effect of each sub-factor for each layer of information expressing the greatest effect of geological factor, land use and then in the following categories of distance from waterway factor: fault and land use for WOE model And then LNRF. Therefore, there is a relatively good relationship between the effective weights identified from the sub-factors in the above models. Evaluation of zonation maps and landslide sensitivity using ROC curve shows that Shannon entropy model with sub-curve surface value (AUC = 0.915) has excellent performance in terms of quality and WOE and LNRF models respectively. With the area below the curve (AUC = 0.899 and (AUC = 0.860)), they have performed well in determining landslide sensitive areas.
The main purpose of this study was to investigate the relationship between factors and sub-factors affecting the occurrence of landslides. The results show that there is a good relationship between Shannon and WOE entropy models and then LNRF in identifying the effective factors based on the weight of the sub-factors. Therefore, in future research, these models can be used well in different conditions to zoning landslide sensitivity. According to the superior model of topographic position index, the factors of distance from waterway, distance from fault and land use have had the greatest effect on the occurrence of landslides. While in comparison, the effect of the weight of the sub-factors is more of the sub-factor that has an effect on the occurrence of landslides than the lithological factor. In general, it can be concluded that the role of natural factors in this basin has been more than human factors and the basin is inherently sensitive to landslides. In most cases, preventing and controlling landslide-prone areas due to natural causes is not possible or will not be easy and, if possible, can be very costly. The high sensitivity of this area to the occurrence of landslides to natural factors causes the least interference to cause a critical situation in the basin. Preparation of landslide susceptibility zoning map and identification of factors involved in controlling and implementing preventive measures in the instability of this basin and formulating a strategic plan - can be very efficient and management of unstable slopes in the study area Facilitate.