پهنه بندي خطر زمين لغزش در حوضه آبريز سيمينه رود بوكان با تلفيق مدلهاي آماري، فرايند تحليل سلسله مراتبي و سامانه اطلاعات مكاني

Landslide Hazard Zoning in the Siminehrood Catchment of Bookan Area by Combining Statistical Models, the Analytic Hierarchy Process and GIS

شناسايي مناطق مستعد وقوع زمين لغزش از طريق پهنه بندي قابليت خطر با مدل هاي آماري مناسب، يكي از اقدامات اوليه در كاهش خسارات احتمالي و مديريت خطر زمين لغزش است. حوضه سيمينه رود از زير حوضه هاي حوضه آبريز درياچه اروميه به دليل موقعيت جغرافيايي و ويژگي‍هاي طبيعي آن يكي از مناطق مستعد كشور براي وقوع زمين لغزش و ساير پديده هاي مرتبط با لغزش دامنه‍ها مي باشد. هدف از انجام اين تحقيق مطالعه و بررسي خطر زمين لغزش در حوضه آبريز سمينه رود مي باشد و براي انجام آن از تلفيق مدل تحليل سلسله مراتبي و مدل آماري وزن هاي شاهد استفاده گرديده است. بطور كلي 9 معيار اصلي مرتبط با وقوع پديده زمين لغزش در اين پژوهش مورد استفاده قرارگرفت. لايه هاي مكاني مربوط به عوامل با جمع آوري، پردازش و طبقه بندي داده هاي مكاني ايجاد گرديدند. لايه وزن طبقات هر لايه توصيف كننده عامل موثر زمين لغزش با استفاده از مدلسازي وزن هاي شاهد توليد شد. وزن اهميت هر لايه با استفاده روش تحليل سلسله مراتبي محاسبه شد. نقشه حساسيت زمين لغزش با همپوشاني لايه هاي وزن طبقات و اعمال وزن اهميت هر لايه ايجاد و طبقه بندي براي حوضه سيمينه رود توليد گرديد. محاسبه ميزان موفقيت نقشه پهنه بندي نشان داد كه حدود %60 زمين لغزش مورد استفاده در تهيه نقشه در طبقه خطر زياد نقشه پهنه بندي قرارگرفته است. تراكم زمين لغزش در اين طبقه حدود 37% مي باشد كه در مقايسه با طبقه خطر متوسط (8.8%) و طبقه خطر كم (0.083%) خيلي زياد مي باشد. ميزان پيش بيني نقشه نيز نشان مي دهد كه حدود %56 زمين لغزش هاي مورد استفاده در ارزيابي نتايج تحقيق در طبقه خطر زياد قرار گرفته اند.

Landslide is a kind of natural disasters that annually causes great losses in life and property throughout the world. With the increasing population and expanding urban areas to the steep slope areas the risk of landslide has increased. The objective of this research is to study the risk of landslide in order to reduce potential damage and manage landslide risk. Siminehrood basin is the case study of the research. It is a subbasin of Orumiyeh lake catchement area at north west of Iran and due to its geographical location and natural characteristics is one of the country's landslide prone areas. In this research nine main factors affecting landslide including distance to fault, distance to road distance to stream network, topography, slope, aspect, landuse, precipitation and lithology are used. The weights of evidence model which is a bivariate statistical analysis and data-driven technique is used for producing Landslide susceptibility mapping. For obtaining the weights of evidence modeling prior probability, conditional probability and the positive and negative weights are calculated and a layer describing the total evidence weight for each class of factor layer is produced. Prior to combining the weights of factor layers, the importance weight of each factor layer is calculated using the Analytic Hierarchy Process (AHP). For implementation of the research the related spatial data were gathered and prepared in order to produce spatial layers for describing effective factors. For preparation of spatial data different processing have been performed such as distance tool for creating distance layer, interpolation tool for creation surface layer from point layer, hydrology tool set for extracting stream network from Digital Elevation Model (DEM) and classification tool. All of the preparation processes, calculation of importance of factor layers by using AHP method and overlaying of factor layers have been implemented in ArcGIS software package. A python application has been developed in ArcGIS environment in order to calculate the importance weight of each factor layer by using AHP. Finally, the landslide susceptibility map of Siminehrood Cachment was produced with overlaying the total evidence weight layers and applying the importance weight of each layer and then the resultant map was classified. For assessing the results of research the total occurred landslide data set in study area was randomly split into two training and testing landslide data sets. Calculation of success rate of landslide susceptibility map using testing landslide data sets indicated that about 60% of landslide pixels which has been used for producing classified landslide susceptibility map are in high risk class and the density of landslide pixels in this class is 37% while in the medium risk class is 8.8% and in the low risk class is 0.083%. Predicting rate of resultant map shows that about 56% of landslides which have been used for verifying and assessing the research results are in the high risk class of classified landslide susceptibility map.