كاربرد سامانه استنتاج فازي براي تخمين عملكرد سيب‌زميني، يونجه و گندم در شهركيان

Application of Fuzzy Inference System to Predict the Yield of Potato, Alfalfa and Wheat in Shahr-e-Kian area

در اين پژوهش كارآيي سامانه استنتاج فازي براي پيش‌بيني عملكرد محصولات مهم شهر كيان، استان چهارمحال و بختياري مورد ارزيابي قرار گرفت و با روش‌هاي مبتني بر منطق ارسطويي بولين (روش سنتي) مقايسه شد. براي اين منظور پس از نمونه برداري از خاك 21 نقطه مشاهداتي با فواصل يك كيلومتر در يك كيلومتر به روش معمول مطالعات خاكشناسي نيمه تفصيلي، انجام تجزيه هاي فيزيكي و شيميايي بر روي آن ها و تعيين ميانگين وزني خصوصيات مزبور تا عمق يك متري خاك، تناسب كمي اراضي به دو روش ريشه دوم (روش سنتي مبتني بر منطق بولين) و سامانه استنتاج فازي (از طريق محاسبه توابع عضويت كلاس هاي تناسب اراضي) براي سه محصول گندم، سيب‌زميني و يونجه تعيين شد. هم‌چنين با استفاده از پارامترهاي اقليمي و بر اساس مدل فائو، توليد پتانسيل محصولات در منطقه مورد مطالعه تعيين شد و در پي آن عملكرد محصول در نقاط نمونه برداري شده، پيش بيني شد.. ﻧﺘﺎﯾﺞ ارزﯾﺎﺑﯽ ﺑﻪدو روش ﺳﺎﻣﺎﻧﻪ اﺳﺘﻨﺘﺎج ﻓﺎزي و روش ﺳﻨﺘﯽ از ﻃﺮﯾﻖ ﻣﻘﺎﯾﺴﻪ ﺿﺮﯾﺐ ﺗﺒﯿﯿﻦ ﻣﻌﺎدﻻت رﮔﺮﺳﯿﻮﻧﯽ ﺑﯿﻦ ﺷﺎﺧﺺ اراﺿﯽ و ﺗﻮﻟﯿﺪ ﻣﺸﺎﻫﺪه ﺷﺪه )واﻗﻌﯽ( در ﺻﺤﺮا ﻣﻘﺎﯾﺴﻪ ﮔﺮدﯾﺪ. ﺿﺮﯾﺐ ﺗﺒﯿﯿﻦ ﻣﺪل ﺑﻮﻟﯿﻦ ﺑﺮاي ﮐﺎرﺑﺮيﻫﺎي ﮔﻨﺪم، ﯾﻮﻧﺠﻪ و ﺳﯿﺐزﻣﯿﻨﯽ ﺑﻪﺗﺮﺗﯿﺐ ﺑﺮاﺑﺮ ﺑﺎ 0/74، 0/6 و 0/64 و ﺑﺮاي ﻣﺪل ﻓﺎزي، ﺑﻪﺗﺮﺗﯿﺐ 0/75، 0/71 و 0/89 ﺑﻪدﺳﺖ آﻣﺪ. ﻧﺘﺎﯾﺞ ﻧﺸﺎن داد ﮐﻪ ارزﯾﺎﺑﯽ ﺗﻨﺎﺳﺐ اراﺿﯽ ﺑﻪروش ﺳﺎﻣﺎﻧﻪ اﺳﺘﻨﺘﺎج ﻓﺎزي، ﮐﺎرآﯾﯽ ﺑﺎﻻﺗﺮي ﻧﺴﺒﺖ ﺑﻪ روش ﻣﺒﺘﻨﯽ ﺑﺮ ﻣﻨﻄﻖ ﺑﻮﻟﯿﻦ در اﯾﻦ ﻣﻨﻄﻘﻪ دارد.

In this study the efficiency of fuzzy inference system to predict the main crops yield of Shahre- Kian area in Chaharmahal and Bakhtiari province was evaluated and compared to the Boolean logic-based land evaluation methods. To do this, soil samples were collected and analyzed physicochemicaly from 21 observation points located on the centers of a 1 km by 1 km grid framework in the study area according to the routine semi-detailed soil survey studies. Weighted average of soil characteristics up to depth of 1 m was used. Then the quantitative land suitability for wheat, alfalfa and potato using square root (conventional Boolean-based method) and fuzzy inference system (by calculating the membership functions of each land suitability class) approaches was determined. Also, the potential yields of the selected crops were calculated based on FAO model using the climatic parameters of the studied area. The expected yield for each observation points was predicted afterward. The accuracy assessment of fuzzy inference system and Boolean methods were carried out by comparing the regression coefficients between calculated land indices and observed (actual) yields in the field. The observed correlation coefficients for wheat, potato and alfalfa were 0.738, 0.642 and 0.6 in Boolean approach, respectively and the coefficients were 0.749, 0.885 and 0.713 in fuzzy inference system, respectively. The results showed that the fuzzy inference system approach had higher efficiency than Boolean method in this area.