تدوین مدل تلفیقی سلول خودکار و زنجیره مارکوف جهت پیش بینی تغییرات پوشش/کاربری سرزمین ( مطالعه موردی: شهرکرج)

ﭘﯿﺶﺑﯿﻨﯽ ﺗﻐﯿﯿﺮات ﭘﻮﺷﺶ/ﮐﺎرﺑﺮی ﺳﺮزﻣﯿﻦ ﻋﻼوه ﺑﺮ درک ﻋﻤﻠﮑﺮد و ﺳﻼﻣﺖ اﮐﻮﺳﯿﺴﺘﻢﻫﺎ، اﻣﮑﺎن ﻣﺪﯾﺮﯾﺖ و ﺑﺮﻧﺎﻣﻪ رﯾﺰی ﺳﺮزﻣﯿﻦ را ﺧﺼﻮﺻﺎ در ﻣﻨﺎﻃﻖ ﺑﺎ ﺗﻐﯿﯿﺮات ﺳﺮﯾﻊ و اﻏﻠﺐ ﻧﺎﻫﻤﺴﻮ ﺑﺎ ﻃﺮحﻫﺎی آﻣﺎﯾﺶ ﺳﺮزﻣﯿﻦ ﻓﺮاﻫﻢ ﻣﯽﺳﺎزد. ﻣﻄﺎﻟﻌﻪ ﺣﺎﺿﺮ ﻋﻼوه ﺑﺮ ﻣﻌﺮﻓﯽ ﻗﺎﺑﻠﯿﺖﻫﺎی ﮔﻮﮔﻞ ارث اﻧﺠﯿﻦ در ﺑﺮداﺷﺖ اﻃﻼﻋﺎت، ﺑﺎ اﺳﺘﻔﺎده از ﻣﺪل ﺗﺮﮐﯿﺒﯽ ﺳﻠﻮل ﺧﻮدﮐﺎر و زﻧﺠﯿﺮه ﻣﺎرﮐﻮف ﺑﻪ ﻣﻘﺎﯾﺴﻪ ﺑﺎ ﻣﺪل ﺗﻠﻔﯿﻘﯽ رﮔﺮﺳﯿﻮن ﻟﺠﺴﺘﯿﮏ، ﺳﻠﻮل ﺧﻮدﮐﺎر و زﻧﺠﯿﺮه ﻣﺎرﮐﻮف ﺟﻬﺖ ﭘﺎﯾﺶ ﺗﻐﯿﯿﺮات ﺳﺮزﻣﯿﻦ و ﭘﯿﺶﺑﯿﻨﯽ اﻟﮕﻮی ﺗﻐﯿﯿﺮات آﺗﯽ ﻣﯽﭘﺮدازد. ﻃﺒﻘﻪﺑﻨﺪی ﺳﻪ ﺗﺼﻮﯾﺮ ﻟﻨﺪﺳﺖ ( 2014 ،2006 و 2021) ﺑﺎ اﺳﺘﻔﺎده از روش ﻣﺎﺷﯿﻦ ﺑﺮدار ﭘﺸﺘﯿﺒﺎن اﻧﺠﺎم ﺷﺪ. ﺑﺎ اﺳﺘﻔﺎده از ﻣﺪل ﺳﻠﻮل ﺧﻮدﮐﺎر و زﻧﺠﯿﺮه ﻣﺎرﮐﻮف و ﻫﻤﭽﻨﯿﻦ ﺑﺎ ﻣﺪﻟﺴﺎزی رﮔﺮﺳﯿﻮن ﻟﺠﺴﺘﯿﮏ ، ﺳﻠﻮل ﺧﻮدﮐﺎر و زﻧﺠﯿﺮه ﻣﺎرﮐﻮف ﺳﺎل 2021 ﭘﯿﺶﺑﯿﻨﯽ ﺷﺪ. ﺟﻬﺖ ارزﯾﺎﺑﯽ دﻗﺖ از ﻧﻘﺸﻪ ﻃﺒﻘﻪﺑﻨﺪی ﺷﺪه ﻫﻤﺎن ﺳﺎل اﺳﺘﻔﺎده ﺷﺪ. دﻗﺖ ﺗﻮاﻓﻖ ﺑﯿﻦ ﻧﻘﺸﻪ ﻫﺎی ﻃﺒﻘﻪﺑﻨﺪی ﺷﺪه و ﻣﺪلﺳﺎزی ﺷﺪه ﺑﺎ اﺳﺘﻔﺎده از رﮔﺮﺳﯿﻮن ﻟﺠﺴﺘﯿﮏ و ﻣﺘﻐﯿﺮﻫﺎی ﮐﻤﮑﯽ ﺑﻪ ﺗﺮﺗﯿﺐ 0/786 =kstandard= 0/757 ،klocation=0/785 ،kno ﻣﯽﺑﺎﺷﺪ. ﭘﯿﺶﺑﯿﻨﯽ ﻣﯿﺰان ﺗﻐﯿﯿﺮات ﺳﺎل 2035 ﺑﺎ اﺳﺘﻔﺎده از ﻧﻘﺸﻪﻫﺎی ﺳﺎلﻫﺎی(2014-2021) و ﺑﺎ ﻣﺪل ﺳﻠﻮل ﺧﻮدﮐﺎر و زﻧﺠﯿﺮه ﻣﺎرﮐﻮف اﻧﺠﺎم ﺷﺪ.ارزﯾﺎﺑﯽ روﻧﺪ ﺗﻐﯿﯿﺮات ﻧﺸﺎن ﻣﯽ دﻫﺪ ﮐﻪ در ﺻﻮرت اداﻣﻪ روﻧﺪ ﻣﺴﺎﺣﺖ ﻃﺒﻘﻪ اﻧﺴﺎن ﺳﺎﺧﺖ از 4839/01 ﻫﮑﺘﺎر ﺑﻪ 7199/76 ﻫﮑﺘﺎر ﺑﺎ 2360/75 ﻫﮑﺘﺎر اﻓﺰاﯾﺶ در ﺳﺎل 2035 ﺧﻮاﻫﺪ رﺳﯿﺪ. اﺳﺘﻔﺎده از ﻣﺪلﻫﺎی ﺷﺒﯿﻪ ﺳﺎزی ﺧﻄﺮﺗﺼﻤﯿﻢ ﮔﯿﺮیﻫﺎ در ﻣﺪﯾﺮﯾﺖ ﺑﻠﻨﺪﻣﺪت ﺳﺮزﻣﯿﻦ راﮐﺎﻫﺶ ﻣﯽدﻫﺪ. ﻫﻤﭽﻨﯿﻦ اﺳﺘﻔﺎده از ﮔﻮﮔﻞ ارث اﻧﺠﯿﻦ ﻣﻮﺟﺐ ﮐﺎﻫﺶ ﻫﺰﯾﻨﻪ و زﻣﺎن ﻃﺒﻘﻪ ﺑﻨﺪی و ﭘﺮدازش ﺗﺼﺎوﯾﺮ ﻣﺎﻫﻮاره ای ﺷﺪه اﺳﺖ.

development of an integrated model based on cellular automata and markov chain to predict land use/cover changes( case study: karaj city)

land cover/use changes detection is very powerful approach to analyze the function and health of ecosystems for land use management, especially in areas with rapid changes and often inconsistent with land use plans. this study has aimed to introduce the capabilities of google earth engine for data gathering. it has implemented the combined model of cellular automata and markov chain in comparison with the integrated model of logistic regression, cellular automata and markov chain to compare the effect of adding auxiliary variables for changes detection and prediction of future changes. three landsat images (2006, 2014 and 2021) were classified using the support vector machine classifier method. with classified maps (2006-2014) and using the cellular automata and markov chain, as well as, adding covariates and modeling with logistic regression, cellular automata and markov chain two land use layers were predicted for 2021. for accuracy evaluation of the simulated layer of 2021, the classified map of the same year was used. the accuracy using logistic regression and covariates was respectively kno=0.786, klocation=0.785, kstandard=0.757. then 2035 was simulated using (2014-2021) layers and the cellular automata and the markov chain. the evaluation of the changes shows that based on bau scenario, between 2006 and 2035, the man-made area will reach from 4839.01 hectares to 7199.76 hectares with an increase of 2360.75 hectares. simulation models will reduce the risks of land use decisions in long-term and google earth engine has reduced the cost and time of classification and processing of satellite images.