How well do the results of conventional land suitability methods fit together?

Choosing the best land suitability method to determine land suitability classes has always been questioned and researched. Many land suitability surveys are conducted using four conventional methods, including: 1-simple limitation, 2- number and intensity of limitations, 3- Khiddir (square root) and 4- storie. In the parametric approach, the land index is calculated using the Khiddir method or the Storie method, and then based on this index, the land suitability classes are determined. The land index is in two forms: the uncorrected land index and the corrected land index. Numerous studies on land suitability evaluation have shown that the results of different land suitability methods are very different. However, the results of many studies suggest that the two methods of simple limitation and Khiddir are more consistent. Since these researches have been for different products and different regions, it has not been possible to evaluate the degree of adaptation of different methods without considering the type of product and its location. On the other hand, the numerical value of the adaptation or conformity of these methods with each other has not been determined. So, regardless of the type of product and location, a basic question that arises is how well do the different land suitability methods fit together, and are the degree of adaptation different for various land suitability classes? Current research seeks to answer this question.

In this study, land suitability classes were determined using numerical simulation, without considering any product and location. In numerical simulation, each of the eight factors involved in determining the suitability classes accounted for a random number from zero to 100, and then based on these simulated numbers the land suitability classes were determined with the four conventional methods and in two forms of the uncorrected land index and the corrected land index. The simulation process for each proportional class, including classes S1(suitable), S2(Moderately suitable), S3(Marginally suitable), N1 (temporary unsuitable), and N2 (permanently unsuitable), was performed one million times. So a total of five million simulations were done. In the next step, the degree of adaptation for each class and each method was calculated based on the Overall Accuracy(OA) formula.

The results showed that the numerical value of the degree of adaptation of different land suitability methods in various classes are very different; Overall, however, the least and most adaptation between the different methods occurs for Class N1 and N2, respectively. After class N2, there is the greatest adaptation between the different methods in class S1. For classes S2 and S3, the degree of adaptation between different methods does not follow a specific trend. Among the various methods, in the case of using the uncorrected index, the simple limitation method was more consistent with the Khiddir method. However, using the corrected land index, the greatest correlation was observed between the simple limitation method and the parametric approach (both Khiddir and Storie method) and in this case there was not much difference between the results of Khiddir method and Storie method. Therefore, the corrected index is superior to the uncorrected index.

Based on the findings of this study, it can be said that the results obtained from the uncorrected land indices may be far from reality, and this issue is much more severe for the Storie method. The use of the uncorrected index causes a large difference between the parametric approach with the two methods of simple limitation and the number and intensity of limitations. Therefore, it is necessary to use corrected indices to determine the suitability classes so that the results of different methods are as close as possible.