Assessment of Corrected Land Index in Land Suitability Evaluation and Adjusting its Functions

In land suitability evaluation using parametric method, Khiddir or square root method (LQSI) and/or Storie method (LSI) are employed to calculate land index (LI), then suitability classes could be determined based on the LI. However, the obtained LI should be corrected according to the minimum rating (Rmin) and then the suitability classes should be determined. The existing functions to correct the LI should be mathematically continuous at all points in order to prevent from losing some LIs and their consequent suitability classes. In the functions represented by Sys, there is a continuity for S1 (suitable), S2 (moderately suitable) and S3 (marginal suitable) classes, but for N (unsuitable) the presented functions are not continuous. Therefore, presented functions for N1 and N2 classes can be very misleading since they are not able to distinguish between N1 and N2 classes and have problem to calculate them.

In this study, the existing functions in the literature were mathematically evaluated for each land suitability classes. Properties and criteria for determining land suitability classes are shown in Table1.  In parametric approach, land index (uncorrected land index) is calculated using Kiddir and Storrie methods as shown in equations 1and 2, respectively. The relationships between uncorrected land indices and corrected land indices are presented in Table 2.   (1) (2)     According to continuity rules, the necessary corrections were made for N1 and N2 classes. Then numerical simulation was employed to assess the obtained results from the both existing and purposed functions and compared them with one another. For this purpose, one million random values were created for each of the S1 to N2 classes; so that the minimum rating (Rmin) was a random number for each class in own defined range and the other seven characteristics were random numbers between Rmin and 100. For example, in the S3 class, a minimum random number is in the range of 40 to 60 and seven other characteristics were between the Rmin and 100. Finally, a total of two million random simulations were created.

Based on the minimum, maximum and mean obtained values the simulation process is acceptable. These numbers show that the simulations have simulated almost all the cases that may occur in reality, from the best to the worst. The results showed that for N1 and N2 classes the correction functions should be respectively 12.5 + 0.314LQSI and 0.5LQSI for the Khiddir method and 12.5+ 0.313LSI and 0.5LSI for the Storie method to maintain the both the continuity of the correction functions for all classes and the corrected land index to be in the defined range for each class. The two million times simulation results also confirmed the accuracy of the obtained functions Therefore, it is suggested to use the proposed functions in determining N1 and N2 classes instead of Sys’s functions.

The use of the usual land index, which is conventionally calculated by the Khiddir or Storie method, called uncorrected land index (UCLI), can be largely misleading without being corrected and converted to the corrected land index (CLI), causing the wrong land suitability classes. Therefore, it is very important to use the relationships that have been developed for this purpose to correct the usual land index. The findings of this study showed that the current functions, although at the order level can distinguish between unsuitable order (N) from the S3 class, but separation between classes N1 and N2 are very difficult to calculate. For this reason, new relationships for N1 and N2 classes were calculated and presented. Therefore, it is suggested that N1 and N2 classes can be used instead of the relationships presented.