مقايسه مدل نسبت فراواني و توابع عضويت فازي در پهنه‌بندي خطر زمين‌لغزش (مطالعه موردي: جاده ارتباطي مريوان- سنندج)

The Comparison of the Frequency ratio model and Fuzzy Membership Functions in Landslide Hazard Zonation (Case study:Marivan - Sanandaj road)

زمين‌لغزش‌ها نوعي حركت دامنه‌اي هستند كه نه تنها سازه‌هاي مهندسي مثل جاده‌ها، خطوط راه‌‌آهن و سدها را تحت تاثير قرار مي‌دهند، بلكه تلفات جاني را هم به دنبال دارند. هدف از اين تحقيق، پهنه‌بندي خطر زمين‌لغزش‌هاي محدوده جاده ارتباطي مريوان - سنندج در استان كردستان (غرب ايران)، از طريق وزن‌دهي به معيارهاي اصلي و با استفاده از مدل نسبت فراواني و توابع ‌عضويت فازي در نرم‌افزار Arc GIS 10 است. به منظور پهنه‌بندي خطر زمين‌لغزش، معيارهاي ارتفاع، شيب، جهت ‌شيب، فاصله از گسل، واحدهاي سنگ‌شناسي، فاصله از آبراهه‌ها، فاصله از جاده، نوع كاربري اراضي، پوشش‌ گياهي، خاك و بارش مورد ارزيابي قرار گرفت. بعد از مرحله فازي‌سازي معيارهاي موثر در وقوع زمين‌لغزش‌هاي منطقه با استفاده از روش‌هاي مذكور، نقشه‌هاي پهنه‌بندي خطر زمين‌لغزش جهت ارزيابي اين دو روش از طريق عملگرهاي گاماي فازي با لانداهاي 7/0، 8/0 و 9/0 تهيه شد و با مقايسه نقشه‌هاي نهايي، مدل ايده‌آل براي پهنه‌بندي خطر زمين‌لغزش‌ در اين محدوده انتخاب شد. نتايج نشان داد كه از ميان دو روش مورد بحث، استفاده از مدل نسبت فراواني به دليل فازي‌سازي مناسب هر معيار با استناد به نقشه‌هاي پراكنش زمين‌لغزش‌هاي رخ داده و نحوه تفكيك طبقات خطر، روش مناسب‌‌تري براي پهنه‌بندي زمين‌لغزش‌ها در اين جاده ارتباطي محسوب مي‌شود.

Introduction The natural phenomenon events are involved with several natural and unnatural factors that play a specific role in it event. Landslides, is example of such natural phenomena that resulted from interaction of various factors. Understanding the impact and weight of each factor by mathematical models can be solution for landslide study in term of zonation and extraction. Landslides are one of the natural hazards which may cause damage to some infrastructures such rail roads, dams, and roads, but also leads to casualties. The landslides hazard Zonation in the study area has been done in this study by using the two models namely frequency ratio model (RF) and functions of the fuzzy model over the Arc GIS 10 software. The final LHZ maps were comparing for evaluated the ideal landslides hazard zonation models. Study Area The study area lied in Marivan – Sanandaj Road with 135 Km length located in longitude of 46?, 07?, 26? to 36? E and latitude of 35?, 23?,11? to 35?, 32?, 35? covered an area of 1281 square kilometers in Kurdistan Province (West of Iran). The study area is part of of Sanandaj - Sirjan techtonic zone in high Zagros range. The 10 years (2000 to 2010) average of annual precipitation in area is around 850 mm/y which is reported by Meteorological Station in Marivan city. Material and Methods In order for landslide inventory map in area, the numbers of variables such as elevation, slope, faults, roads and drainage buffers, land use/ land cover, soil type, annual rainfall, as well as rock type in from of digital maps and layers has been input in a GIS geo data base. The necessary information’s were obtained from several sources such as geology and topography maps, ETM satellite images etc. These data were weighted and analyses by both the Frequency Ratio (RF) and Fuzzy Membership Functions (FMF) models over the Arc GIS environment. The reference landslide map was provided from Iranian Forests and Rangelands bureau. Landslide Hazard Zonation map was done based on two mentions methods. In FR method, weighting of the criteria was done by combining the landslides distribution map with the individual field measurements. In second method we determine the weights of criteria using fuzzy membership functions. The weight outcome classes of each criterion’s in a both methods were normalized (between zero and one), where the Fuzzy Membership Functions and frequency ratio equation used for standardization over the all prepared raster maps. The final LHZ maps were superimposed and different fuzzy Gama operators applied for map accuracy evaluation. The result has shown the LHZ class’s outcome from FR method in more precision than FMF with gamma operator result of 0.7, 0.8 and 0.9. Results and Discussion The results showed that most of the landslide occurred in an areas with slope of more than 70 percent especially in slop with azimuth 0 to 90 (north and northeast slopes). It is also observed that the area with elevation in range of 1049-1350 m and shallow soil over rock as well as irrigated cultivate land are more sensitive to the land sliding. In terms of distance from other phenomena, the 100 meter radius from roads and faults were the most susceptible sloped region for landslides. The two main evaluation faction applied methods is: A - Evaluation of RF method and FMF for standardization of criteria apply for the landslides occurrence. The study outcome showed that LHZ using the standardized with frequency ratio model is more precision rather than other. This is because of fuzzy weight extraction of for each criteria class in sliding or non-sliding accordance pixels and frequency ratio equation applied. B - Evaluation of RF method and FMF for preparation of the LHZ maps. The fuzzy gamma operator was applied over the superimposing standard landslides hazard zonation maps of the area that prepared by two methods. The high precision LHZ map was selected as amp with greater percentage of the sliding pixels. This study showed that the accuracy of hazard zonation maps for sliding processes was directly associated with important steps of criteria class standardization and methods selected. Out of two discussed methods, the FR method was marked as suitable method for LHZ in term of criteria class standardization and landslides occurrence pixels. In an area such as study area (Marivan – Sanandaj Road) Conclusion The Marivan – Sanandaj Road have importance role in international transit and markets access in the western part of Iran. In this study, the weighting of the criteria for the occurrence of landslide were compared by frequency ratio method and fuzzy membership functions. The outcome result based on spatial and statistical analysis was showed frequency ratio method as precision method. This because of high capability of this method for criteria class extraction and standardization over the land sliding input map as well as reality in fuzzy functional.