Optimizing the fuzzy model for land suitability evaluation by GA approach and choosing the kind of membership function

The land suitability analysis can be the first prerequisite in sustainable agriculture. A model was designed to determine the soil index based on fuzzy approach. The study area was parts of Abyek and Nazarabad towns in Qazvin and Alborz province of Iran, respectively. The used fuzzy inference system was Mamdani type. In this fuzzy model, the eight number of soil characteristics were selected to assess the land suitability for winter wheat production. The output of the model was the soil index as a number between zero and 100. Fuzzy rules were written using knowledge of some experts. GA approach was used as a complementation of fuzzy approach to optimize fuzzy rules of the model. The results showed that the value of determination coefficient for model after optimization was 0.952 and significant at the 1% level. The results showed the accuracy of fuzzy-GA model and the power of fuzzy method in the modeling. In the other hand, choosing the kind of membership function had a significant effect on result. So to find the best combination of membership function, a total of 24 different combinations of membership functions were tested for fuzzy sets of inputs. The best combination of membership function was triangular for the medial fuzzy sets and trapezoidal shape for the side fuzzy sets respectively. Classification maps for soil units well prepared in the fuzzy-GA method.

Keywords:

Fuzzy , GA , membership function , Soil index