نوع مقاله : مقاله پژوهشی

نویسندگان

1 دانشکده کشاورزی و منابع طبیعی داراب، دانشگاه شیراز، ایران

2 ژئومورفولوژی بخش جغرافیا، دانشکده ادبیات و علوم انسانی، دانشگاه شیراز، ایران

چکیده

امروزه شناسایی لندفرم­ها و طبقه‌بندی زمین مبتنی بر روش کارشناسی می باشد که به صورت دستی و با استفاده از عکس های هوایی و نقشه های توپوگرافی انجام می شود که روشی وقت گیر و دارای دقت کمی می باشد. از این رو استفاده از روش های نیمه اتوماتیک و اتوماتیک به منظور  طبقه بندی لندفرم ها برای افزایش دقت و سرعت کار، ضروری به نظر می رسد. این پژوهش سعی دارد که به  طبقه بندی لندفرم ها بر اساس الگوریتم شبکه های عصبی خودسازمانده (SOM)در حوضه آبخیز گاوخونی بپردازد. پژوهش از نوع تحلیل و توصیفی مبتنی بر روشهای آماری، نرم افزار و میدانی است بدین صورت که که به منظور استفاده از الگوریتم SOM برای طبقه بندی لندفرم ها از 6 پارامتر استفاده شد که  شامل جهت (aspect)، ارتفاع (elevation)، شیب (slope)، پروفیل طولی و عرضی (plan , profile) و انحنا (curvature) می باشد. برای این منظور ابتدا با استفاده از شاخص موقعیت توپوگرافی (TPI)، لندفرم های منطقه مورد مطالعه طبقه بندی شدند که  از کلاس های لندفرم حاصل از TPI به منظور آموزش مدل SOM استفاده شد. در مرحله بعد از 50 نقطه به عنوان نمونه برای آموزش شبکه استفاده گردید. نتایج حاصل از طبقه بندی لندفرم ها با استفاده از الگوریتم SOM نشان داد که 6 خوشه (کلاس) در محدوده مورد مطالعه وجود دارد، به طوریکه خوشه 1 و 5 شامل لندفرم هایی است که در ارتفاعات زیاد قرار دارند و  خوشه 3 شامل لندفرم هایی است که در کمترین ارتفاع واقع شده اند. بقیه خوشه ها شامل لندفرم هایی هستند که در ارتفاعات متوسط در حوضه آبخیز مورد مطالعه قرار دارند. بنابراین از الگوریتم فوق می توان به منظور پیش بینی لندفرم های منطقه مورد مطالعه استفاده کرد.

کلیدواژه‌ها

موضوعات

عنوان مقاله [English]

Landformclassificationusingself-organizingneural networks(Self-organization map)(Case Study:BasinGavkhoni)

نویسندگان [English]

  • Marzieh Mokaram 1
  • Saeed Negahban 2

1 Darab Faculty of Agriculture and Natural Resources, Shiraz University, Iran

2 Geomorphology, Department of Geography, Faculty of Literature and Human Sciences, Shiraz University, Iran

چکیده [English]

system of landform classification for soil mapping has been desired by soil scientists in Canada for a long time. The Canada Soil Survey Committee (CSSC) adopted a system at a meeting held at the University of Guelph in February 1976. Many aspects of the system came from mapping schemes used by the Geological Survey of Canada for mapping surficial geology. The system also embodies concepts developed initially by R.J. Fulton and later by N.F. Alley while doing terrain mapping in British Columbia. However, the needs of the soil scientist for a terrain or landform classification system are not necessarily compatible with those of the geologist.  Relief analysis is a tool to analyse a landscape based on a Digital Elevation Model (DEM). One of the simplest parameters might be the elevation itself, or slope or the exposition of a given point in a landscape. Moore et al. (1991) state that the spatial distribution of topographic attributes can often be used as an indirect measure of the spatial variability of hydrological, geomorphologic and biological processes. The advantage compared to other information such as soil parameters or biomass production estimates is based on the relatively simple and fast techniques to model processes in large areas and the complex spatial patterns of environmental systems as seen by Moore et al. (1993b). Another relief parameter relevant for this work is landforms or relief units. Each of these contains certain characteristic physical, chemical, and biological processes and parameters (see Dehn et al., 2001). Milne (1936) was one of the first scientists, who recognised the catena principle of soil formation in a hilly terrain (Ruhe, 1960).
Material And Methods
Materials are classified according to their essential properties within a general framework of their mode of formation. Four groups (components) of materials have been recognized to facilitate further characterization of the texture and the surface expression of the materials. They are unconsolidated mineral, organic, consolidated, and ice components. These groups and the classes established within them are presented below. This research is trying to classify landforms on the basis of self-organizing neural network algorithm (SOM) in the watershed Gavkhoni pay to use the SOM algorithm is used to classify landforms of 6 parameters that includes orientation (aspect), height (elevation), tilt (slope), the longitudinal and transverse profiles (plan, profile) and curvature (curvature) is. Generally, The aim of this studyis the classification of landformsin the basin Gavkhoni. Classification methods to help major landforms visit the field, using topographic maps and aerial photos, which requires experience. The automatic method based on digital elevation model(DEM)can be used to classify landforms Basin Gavkhoni.
Result And discussion
The results of the classification of landforms using SOM algorithm showed that 6 cluster (class) in the study area there as clusters 1 and 5 includes landforms that are at high altitudes and cluster 3 includes landforms that are located at the lowest height. The rest of the cluster, including the landforms that the average height of the watershed studied. So the algorithm can be used to predict the landforms of the study area.
The results showed that6isthe maximum data in SOM algorithm. Also, at leastin this hex data is zero, which indicates that there are no number sinthislocation. The results of principal component analysis showed high density and distribution data. According to the above results show that the landforms input data in Figure6 classes have been distributed in the study area.
 
Conclusion
In this research was used SOM (SOM) to classify landforms. In order to use algorithms for classification of landforms of 6 parameters were used in the watershed Gavkhoni , The results of the classification of landforms using SOM algorithm showed that 6 cluster (class) in the study area ther. as clusters 1 and 5 includes landforms that are at high altitudes and cluster 3 includes landforms that are located at the lowest height. While cluster 3 includes landforms that are the lowest height. The rest of the cluster, including the landforms that the average height of the watershed studied. In general, using the SOM algorithm can be 6 classes to classify landforms in the study area predicted.  Using the results of the SOM algorithm to manage watershed management approaches should be considered 6.

کلیدواژه‌ها [English]

  • Classification of landforms
  • self-organizing neural networks (SOM)
  • topographic position index (TPI)
  • Gavkhoni Basin