پیوندهای مفید
اخبار و اعلانات
آمار
| تعداد نشریات | 161 |
| تعداد شماره‌ها | 6,573 |
| تعداد مقالات | 71,037 |
| تعداد مشاهده مقاله | 125,524,628 |
| تعداد دریافت فایل اصل مقاله | 98,785,231 |
کاربرد الگوریتم‌های طبقه بندی در پهنه بندی خطر فرسایش بادی استان اصفهان | ||
| پژوهش های جغرافیای طبیعی | ||
| دوره 56، شماره 3، آبان 1403، صفحه 59-75 اصل مقاله (1.16 M) | ||
| نوع مقاله: مقاله کامل | ||
| شناسه دیجیتال (DOI): 10.22059/jphgr.2024.382042.1007839 | ||
| نویسنده | ||
| احمد حجاریان* | ||
| گروه برنامه‌ریزی روستایی، دانشکده علوم جغرافیایی و برنامه‌ریزی، دانشگاه اصفهان، اصفهان، ایران | ||
| چکیده | ||
| فرسایش خاک ممکن است به از بین رفتن کیفیت خاک و در نتیجه، کاهش بهره‌وری خاک منجر شود. یکی از راه‌کارهای اساسی برای مهار و کاهش اثرات مخرب فرسایش، شناسایی مناطق مستعد فرسایش در مناطق است. به همین دلیل تعیین مناطق مستعد به فرسایش نقش مهمی در مدیریت فرسایش در منابع طبیعی دارد. تحقیق حاضر درصدد است تا با استفاده از دو روش یادگیری ماشینی جنگل تصادفی و ماشین بردار پشتیبان و 3327 نقطه وقوع فرسایش، مناطق مستعد به فرسایش را در استان اصفهان را تعیین نماید. عوامل محیطی در چهار گروه اصلی شامل عوامل توپوگرافی، عوامل اقلیمی، عوامل زیستی و عوامل انسان‌ساخت تهیه شدند. بررسی شاخص AUC نشان داد که هر دو مدل دارای دقت مناسبی بوده هرچند مدل جنگل تصادفی (AUC = 0.97) دارای دقت بالاتری نسبت به مدل ماشین بردار پشتیبان (AUC = 0.86) بود. بر اساس نتایج مدل ماشین بردار پشتیبان، حدود 6/11 درصد در کلاس پرخطر و حدود 6/68 درصد در کلاس خطر کم فرسایش قرار دارد. همچنین در مدل جنگل تصادفی حدود 1/20 درصد در کلاس پرخطر و حدود 2/49 درصد در کلاس خطر کم قرار دارد. در این زمینه، نتایج به‌دست‌آمده می‌تواند با ارائه گستره‌ای از مناطق استان اصفهان و بهره‌گیری از این دو مدل، به تصمیم گیران و برنامه‌ریزان در تدوین برنامه‌ای مدیریتی و نیز اتخاذ اقداماتی راهبردی در جهت کنترل فرسایش کمک کند. | ||
| کلیدواژه‌ها | ||
| استان اصفهان؛ جنگل تصادفی؛ فرسایش خاک؛ مدل‌سازی؛ ماشین بردار پشتیبان | ||
| عنوان مقاله [English] | ||
| The use of classification algorithms in wind erosion risk zoning in Isfahan province | ||
| نویسندگان [English] | ||
| Ahmad hajarian | ||
| Department of Rural Planning, Faculty of Geographical Sciences and Planning, University of isfahan, Isfahan, Iran | ||
| چکیده [English] | ||
| ABSTRACT Urban weathering refers to the processes of loosening, decay, and eventual deterioration of materials used in various urban constructions. This study focuses on assessing the degree of weathering in gravestones from the Joy-e-Horhor and Khold-e-Barin cemeteries in Yazd. A combination of petrographic analyses and longitudinal monitoring of Schmidt hammer rebound values for hundreds of gravestones was employed to achieve this. The findings indicate that in addition to the petrographic characteristics of the stones, such as mineralogical composition, mineral diversity, and rock texture and fabric, local climatic conditions significantly influence the weathering and degradation of these materials. Key processes contributing to the loss of stone durability include temperature fluctuations leading to thermal expansion and contraction, the albedo effect of the stone, wet-dry cycling, and the crystallization and dissolution of secondary minerals like calcite and gypsum. Gravestones made of travertine and marble, characterized by a predominance of calcite minerals and light-colored surfaces, exhibit higher resistance to weathering compared to other lithologies, provided they are not exposed to excessive moisture or frequent washing. In contrast, low-grade metamorphic rocks such as slate and phyllite are the least suitable for gravestones due to their high density of fractures and cleavage planes. Similarly, dark-colored igneous rocks are prone to rapid durability loss, as the differential thermal expansion and contraction of their constituent minerals in response to temperature changes accelerate their weathering processes. Extended abstract Introduction The role of humans in soil erosion is significant directly and indirectly. For this reason, severe damage caused by soil erosion has prompted managers and researchers to make great efforts to predict and spread soil erosion. Iran is one of the countries that has severe soil erosion due to its location in the dry and semi-arid belt. Examining the surface of the areas affected by soil erosion shows that soil erosion has an increasing trend, so that the survey of land area shows an increasing trend in these areas. Isfahan province has the necessary and high potential for soil erosion due to the special conditions of the region, such as topography, slope, lithological condition (presence of formations with low permeability) and climatic conditions (arid and semi-arid). Therefore, since soil erosion is accompanied by numerous damages, it is very important to check the points prone to erosion risk. According to the mentioned points, the phenomenon of erosion can be investigated and studied despite all its complications. One of the management methods of dealing with soil erosion is determining the points prone to erosion. Therefore, the purpose of this research is to evaluate and determine areas prone to erosion using machine learning methods in Isfahan province. Methodology In this research, the main goal is to determine the areas prone to erosion. For this reason, in the first step, based on past research, as well as experts' opinions and examining the history of erosion areas, the factors affecting the occurrence of erosion were first determined. Among the influential factors, we can mention 4 main criteria (topographic, biological, climatic and man-made factors) and 17 sub-criteria. Each criterion and its sub-criteria are briefly described below. The map related to all environmental factors was created in a raster format with a pixel size of 30 x 30 square meters. After preparing layers of environmental factors, information about erosion points was extracted. In this research, all information related to the occurrence of erosion in the last 20 years was collected from available sources. In this regard, in addition to the information of the departments of natural resources, watershed management and environmental protection, it was used. In total, after removing duplicate points and merging points with a distance of less than 100 meters, the final layer of erosion points was created as a point. This information was divided into two groups of points necessary for training the model (70%) and points necessary for evaluating the results of the model (30%) completely randomly. In total, 2543 points were used to train and build the models and 784 points were used to evaluate the accuracy of the models. The modeling related to two models of random forest and support vector mation was done in R software and "randomForest" and "e1071" packages, respectively. Also, to determine the relative importance of environmental variables in the occurrence of erosion in Isfahan province, the method of relative importance diagram (varImpPlot) was used in the random forest algorithm. Results and discussion Investigating the relative importance of the layers used in modeling using the random forest algorithm showed that the layers of vegetation and precipitation were the most important factors in the occurrence of erosion in Isfahan province, respectively (Figure 3). Also, the distance layer from the residential areas was the least important factor in the areas with a history of erosion in Isfahan province. According to the results, in the support vector machine model, the low risk class had the largest area of 7345617 (68.6 percent). In the support vector machine model, the second class with a large area was related to the medium risk class, which includes 19.8% of the study area. While in the random forest model, the average class with an area of 3,276,567 thousand hectares, equivalent to 30.6 of the area, has more value than the support vector machine model. Finally, in the support vector machine model, the lowest area was related to the high-risk class, which covered only 11.6% of the area, while in the random forest model, the lowest area was related to the high-risk class with an area equal to 2157652 hectares. (20.1 percent) was. The evaluation of the accuracy of the support vector machine and random forest models based on the area under the graph (AUC) in the ROC curve showed that both models have good accuracy, but the random forest model has a higher accuracy. Based on the results, the random forest model had a level under the graph of 0.97 and the support vector machine model had a level under the graph of 0.86. Conclusion Modeling of erosion-prone areas can be a key tool for correct and timely management of natural hazards, including soil destruction. Considering that Iran is located in the dry and semi-arid belt, soil erosion is one of its most important crises. Many managers are trying to discover suitable solutions for erosion management that can be used to manage and control erosion in the shortest possible time. Considering the time required for erosion control and resource investment, accurate estimation of the risk of erosion and preparation of erosion distribution maps is the first step in erosion management and risk assessment. The present research was also carried out in order to investigate the efficiency of different machine learning methods in modeling areas prone to erosion in Isfahan province. In the present research, based on the results of modeling, the layers of vegetation and precipitation were the most important factors in the occurrence of erosion in Isfahan province, and the distance from residential areas was the least important factor in areas with a history of erosion in Isfahan province. Based on the conducted studies, temperature, air humidity, and increase in rainfall are among the natural factors that provide the basis for extensive surface erosion in the regions. In general, the accuracy of the used model depends on many factors, which can be attributed to the characteristics of the studied area, such as topography, factors affecting the occurrence of erosion, the accuracy and type of layers of independent variables for modeling the probability of the risk of erosion, the accuracy of points And the recorded ranges of erosion pointed to the continuous occurrence in the past and the type of prediction algorithm used among the factors affecting the overall accuracy of the classification. Considering the increase of erosion rate and increase of crisis in the regions, the results of this research can be used as a model for erosion management in the study area and help managers in planning and actions and providing facilities in high-risk areas and prone to erosion. Funding There is no funding support. Authors’ Contribution All of the authors approved the content of the manuscript and agreed on all aspects of the work. Conflict of Interest Authors declared no conflict of interest. Acknowledgments We are grateful to all the scientific consultants of this paper. | ||
| کلیدواژه‌ها [English] | ||
| Isfahan Province, Random Forest, Modeling, Soil Erosion, Support Vector Machine | ||
| مراجع | ||
| ||
|
آمار تعداد مشاهده مقاله: 48 تعداد دریافت فایل اصل مقاله: 43 |
||
سامانه مدیریت نشریات علمی. قدرت گرفته از سیناوب