تغییرات کاربری اراضی و شبیه سازی رشد و توسعه شهری رشت با استفاده از مدل شبکه عصبی و سلولهای خودکار زنجیره مارکوف
محورهای موضوعی : فصلنامه علمی و پژوهشی پژوهش و برنامه ریزی شهری
داود اکبری
1
(گروه نقشه برداری، دانشکده مهندسی، دانشگاه زابل، زابل، ایران)
مینا مرادی زاده
2
(گروه مهندسی نقشه برداری، دانشکده عمران و حمل و نقل، دانشگاه اصفهان)
محمد اکبری
3
(گروه مهندسی عمران، دانشکده فنی، دانشگاه بیرجند)
کلید واژه: کاربری اراضی, تصاویر ماهوارهای, شهر رشت, مدل شبکه عصبی, سلولهای خودکار زنجیره مارکوف,
چکیده مقاله :
امروزه با توجه به روند رو به رشد جمعیت در شهرها و روستاها و تمایل به زندگی شهری بیشترین تغییرات کاربری اراضی در نواحی شهری به وقوع میپیوندد. توسعه سریع شهری در دهههای اخیر موجب تغییرات وسیعی در الگوی کاربری زمین پیرامون شهرها شده و تأثیرات زیستمحیطی و اجتماعی–اقتصادی زیادی را به همراه داشته است. در این پژوهش با استفاده از تصاویر ماهوارهای تغییرات کاربری اراضی و شبیهسازی رشد و توسعه شهر رشت به کمک مدل شبکه عصبی و سلولهای خودکار زنجیره مارکوف انجام گرفت. برای این منظور از تصاویر سالهای 2000، 2008 و 2017 ماهواره لندست استفاده گردید. پس از پیش پردازش تصویر و انتخاب بهترین ترکیب باندی، تصاویر با روش شبکه عصبی طبقهبندی شد. سپس تصاویر طبقهبندی شده در مدلساز تغییرات زمین وارد گردید و نقشههای خروجی مدلساز با روش CA-MARCOVE برای سال 2027 پیشبینی شد. نتایج بدست آمده در فاصله زمانی 2000 تا 2017 نشان میدهد که تغییرات مساحت در اراضی شهری، شالیزار و جنگل به ترتیب بهمیزان 87/9041، 03/7841 و 78/55 هکتار بوده که میزان آن در شهر رشت مثبت و در شالیزار و جنگل منفی میباشد و نقشه پیشبینی سال 2027 با روش CA-MARCOVE نیز موید افزایش قابل توجه کاربری شهری به مقدار 04/14105هکتار در سالهای آتی است. نتایج این پژوهش نشان می‌دهند که ادامه روند فعلی تغییرات کاربری اراضی به نتایج مضر زیستمحیطی و به‌تبع آن آسیب‌های اقتصادی- اجتماعی جبران‌ناپذیر می‌انجامد. بنابراین ضروری است دستگاه برنامه‌ریزی و مدیریت منطقه، رویکردی جامع برای جلوگیری از مشکلات زیستمحیطی آتی و مهار توسعه افقی سکونتگاه‌ها در این منطقه در پیش گیرد.
Abstract
Nowadays, most land use changes occur in urban areas, due to the growing population in cities and villages and the desire to live in urban areas. Urban rapid development in recent decades has led to large changes in the cities around and has had many environmental impacts. In this research, we evaluated land use changes and urban development simulation using satellite imagery and with neural network model and Markov chain auto-cells in Rasht city. For this purpose, Landsat satellite imageries were used from 2000, 2008 and 2017. After preprocessing the image and selecting the best band combination, the images were classified using the neural network method. Then the classified images were entered into the land changes model and predicted modeling output maps using the CA-MARCOVE method for 2027. The results obtained between 2000 and 2017 indicate that the area changes in urban lands, rice fields and forests were 9041.88, 7841.33 and 55.78 hectares, respectively, which were positive in Rasht city and negative in rice fields and forest and the projection map for 2027 with the CA-MARCOVE method also indicated a significant increase in urban use of 14105.04 hectares in the coming years. The results of this study indicate that the current trend of land use changes will lead to adverse environmental impacts and, consequently, irreversible socio-economic damage. Therefore, it is essential for the region planning and management unit to adopt a comprehensive approach to conduct future environmental problems and to curb the horizontal development of settlements in the area.
Key words: Land Use, Satellite image, Neural network model, Markov chain auto-cells, Rasht city.
Extended Abstract
Introduction:
The importance of land use as a key component in natural resource management, environmental change and a dynamic and affecting biological condition requires accurate quantitative and qualitative information to be provided and varied in the short term. (Triantakonstantis & Stathakis, 2015: 194; Akbari and Rezaei, 1397: 94). In the meantime, remote sensing data provide valuable multi-temporal data on the processes and patterns of land cover change and land use, and help to develop an understanding of the impact of human activities on natural resources. (Esfahanzadeh, 2016: 34). Urban development is a global phenomenon and one of the most important phenomena that has a great impact on both nature and human environment due to its many ecological and socio-economic aspects. The city of Rasht, like other urban areas, has undergone numerous changes in agricultural and horticultural uses and residential uses over many years. In this study, satellite imagery is used to evaluate land use changes and simulate urban development in the period 2000 to 2017, so that the results of the research can be of great help in micro planning. And provided the experts with a great deal to prevent environmental degradation.
Methodology:
In this study, using satellite imagery of land use changes and simulation of growth and development of Rasht city using neural network model and Markov chain automated cells. Landsat 2000, 2008 and 2017 images were used for this purpose. After image preprocessing and selecting the best band composition, the images were classified by neural network method. Selected classes include 7 classes, forest, man-made areas, paddy fields, sand, sea, ponds and vacant lots. The digital layers used to classify and apply Markov auto cells include: GPS capture points for image classification and accuracy assessment, proximity to main roads, river avoidance, distance from surrounding villages, slope And height. Then, the classified images were entered into the land change modeler and the model outputs were predicted by CA-MARCOVE for 2027.
Results and discussion:
The results show that out of the total area of man-made area increased, 3612 hectares were converted to paddy fields and 1 hectare to water use, 2138 hectares were made to man-made areas, 1646 hectares to the sea and 24 hectares to the Bayer area. In the present study, Markov chains and automated cell fusion methods were used to predict land use changes in Rasht. To do this using IDRISI software, three series of land use maps were prepared for the years 2000 to 2017. Finally, based on the factors involved in urban land use changes in the study area, the inputs of the automated cell model were selected as Table (1). The prediction is a function of the model inputs.
Table 1: Input variables in the automated cell simulation model
Row
Variables affecting land use
1
near the main ways
2
distance from the river
3
elevation
4
slopes
5
distance from surrounding villages
Source: Authors' Studies, 2018
Then, by calculating the Kramer coefficient in the model, one can obtain an estimate of the correlation of each variable with the existing land uses and hence its ability to predict land use changes. By repeating 10,000 times of trial and error in the multilayer neural network, calibration and conversion potential maps were generated in the images from 2000 to 2008 and 2008 to 2017. Following the acceptable accuracy of the model for prediction, using the CA-Markov model, the 2027 User Prediction Map was prepared for the study area shown in Figure (1).
Figure 1: Land use forecasting map of Rasht city using CA-Markov for 2027, Source: Research Findings, 2018.
Table 2. Land use area of 2027 using CA-Markov
User class
2027 forecast area (ha)
Jungle
0/13
Sands
0/9639
Water
974/26
rice field
4797/82
Man-made areas
14105/04
Wasteland
1599/03
Sea
6/39
As can be seen from Table 2, the area of most land uses, except for man-made areas and the sea, declined as forest land use from 1031/95, sandy land from 15/42, water from 22/333, paddy fields from 88/12/85 and wasteland from / 66. 3629 hectares decreased in 2000 to 0.13, 0.96, 97.26, 47.72 and 15.03 ha in 2027, respectively. In contrast, the land use area of the man-made areas increased sharply to 14105.04 hectares, while sea use increased by 6.39 hectares.
Conclusion:
The use of Landsat satellite imagery is useful in terms of availability, duplicate coverage and lower cost of source data, as well as determining the extent of land cover changes and land use prediction using the models used in Research can be a good alternative to costly methods of discovering change in the shortest time possible. Other objectives of this study were to use satellite imagery and LCM tools to detect changes occurring in the region during the study years 2000–2008 and 2008–2017. Therefore, multi-layer neural network method was used to detect the changes. Examination of changes from 2000 to 2008 showed an increase in urban class area, with the city area increasing from 6793.91 hectares in 2000 to 8940.41 hectares in 2008. The highest increase in urban area was observed from 2008 to 2017 after image classification. During the study periods, paddy, forest and wilderness land use has been steadily declining, and vegetation use has had a protective role as urban land use. In this study, the prediction of physical growth in the city of Rasht in the coming years (2027) was investigated. This is how the 2017 forecast map was first derived using the CA-MARKOV model. Comparison of the results of the prediction map with that of the image classification showed high accuracy. The 2027 forecast map also shows a significant increase in urban land use by 14105.04 hectares in the coming years. Considering the results, it is possible to study changes in vegetation cover and to prevent its unnecessary changes and transformations. Because vegetation plays an important role in reducing environmental issues in urban areas. In contrast, the disappearance of vegetation causes severe environmental crises in relation to the rapid growth of urbanization and the formation of the thermal island of the city. As a result, vegetation is considered as an indicator of environmental sustainability in urban communities. Therefore, proper vegetation management is considered as an integral part of any sustainable urban development. Since degradation of vegetation and rising surface temperature can have adverse effects on the environment, identifying environmental sensitivities (crises) caused by this factor is essential as it can play an important role in urban development management.
Ahadnezhad, M. and S.A. Hosseini, (2011): The Evaluation and Predicting of Urban Sprawl using Multi-Temporal Satellite Imagery and GIS (A Case Study Tabriz City 1984-2010). Research and Urban Planning, Vol. 2 , No. 4., pp. 1-20.
2. Aithal, B.H., V. Raok, and T.V. Ramachandra, (2014): Modeling and geo-visuisation of urban growth, Conference on conservation and sustainable management of wetland ecosystem in Western Ghats.
3. Akbari, M. and M.R. Rezaei, (2018): Assessment of Land Use Changes in the 3rd Zone of Isfahan Metropolis. Research and Urban Planning, Vol. 9, 34, pp. 93-104.
4. Ashari, E., A. Abdollahi, S. Karimzade and R. Azizi (2015): Study on urban land use change impacts on rural settlement strategy using automated cell model (case study city of Urmia). Journal of Geographical Space Preparation, Golestan University Scientific-Research Quarterly, pp. 161-151.
5. Dadashpoor, H. and A. Zarei, (2013): Forecasting of Urban Development Changes in Noshahr County Using Lucia Model. Research and Urban Planning, Vol. 3 , No. 11, pp. 37- 58.
6. Esfahanizadeh, R. (2016): Urban Development Analysis Using Remote Sensing and Spatial Information System, Case Study: Lahijan City,MSc Thesis, Department of Civil Engineering, Department of Civil Engineering, Assistant Professor : Dr. Rahmatullah Farhudi.
7. Ganeshkumar B. and M. Mohan, (2014): Urban Sprawl Spatial Modeling using SLEUTH Model. International Journal of Geospatial Engineering and Technology, Vol. 1, No. 1, pp. 22 – 28.
8. Ma'boodi, M., T. Hakimi and H. Hakimi (2015): Analysis of Land Use Changes and Urban Sprawl Simulation in Mid-Sized Cities (Case Study: Khoy City). Urban Planning Geography Research, Vol. 3, No. 2, pp. 221-222.
9. Nazemfar, H. and J. Jafarzadeh (2019): Investigation of the Effectiveness of Satellite Image Classification Methods in Evaluating Urban Land Use Changes Using Scale Optimization in Object Oriented Processing (Case Study: Ardabil City). Journal of Research and Programming Urban Planning, Vol. 10, No. 36, pp. 117–128.
10. Richard, J.A. and X. jia, (2006): Rmeote sensing digital image analysis: An introduction 4th, springer, 439p.
11. Singh, A. (1989): Digital Change Detection Techniques Using Remotely Sensed Data. International Journal of Remote Sensing, 10(6), pp. 989-1003.
12. Triantakonstantis, D. and D. Stathakis, (2015): Urban Growth Prediction in Athens, Greece, Using Artificial Neural Networks. International Journal Civil structural Constr. Archit. Eng, 9, 193-197.
13. Vafaie, S., A.A. Darvishat and M. Pirboghar (2013): Monitoring and Forecasting of Land Use Spatial Changes Using the LCM Model (Case Study: Marivan Area). Iranian Forestry Journal, Iranian Forestry Association, Vol. 5, No. 3, pp. 323-336.
14. Wang, S., Q. Zheng, and X. Zang, (2012): Accuracy assessments of land use change simulation based on Markov-cellular automata model. Procedia Environmental Sciences 13:1238–1245.
15. Xian, G. and M. Crane, (2014): Assessments of urbangrowth in the Tampa Bay watershed using remote sensing data. Journal of Remote Sensing of Environment. No 97.pp. 203-215.
16. Yuan, D., C.D. Elvidge, and R.S. Linetta, (2002): Survey of multi spectral methods for land cover change analysis. Remote Sensing Change Detection environmental monitoring methods and application, 43, pp. 24-33.
17. Zarabi, A., J. Alizadeh and H. Askari, (2012): Analysis and Evaluation of Land Use in Small Rural-Towns using GIS Case Study: Rural-Town Of Aslandvz (In Ardebil Province). Research and Urban Planning, pp. 51-76.
_||_