Where Ni is the standardized value of variable i, Xi is the original data of I, and min and max are the minimum and maximum values of the original data set of i.
Binary response data ([0,1]) were assigned to the categorical variables (Table 1). Land use was recoded into a binary map as described by Hu and Lo (2007). Eleven variables(ST1-ST11), 5 variables (PM1-PM5), 4 variables (OM1-OM4), and 6 variables (LU1-LU6) were generated to distinguish 12 soil types, 6 parent material types, 5 organic content grades, and 7 land use types, respectively.
All of the explanatory variables were converted into a rester format with a 30-m resolution for the simulation. The spatial distributions of explanatory variables are shown in Fig. 3. The categorical variable maps (Fig. 3m-p) were reclassified into binary maps in ArcGIS 10.0 before the simulation.