Evaluating the effectiveness of environmental variables in modeling spatial distribution of cation exchange capacity in Sistan Plain using geostatistical approaches

Background and Objectives: Soil maps are essential for various users and land-use decision-makers. Improving the accuracy of these maps is of significant importance. Soil properties exhibit complex spatial and temporal variations due to the combined interactions of biological, physical, and chemical processes at different scales. Precise knowledge of the spatial distribution of soil characteristics is crucial for effective agricultural management and informed decision-making. Soil cation exchange capacity (CEC) is an important property that serves as a reliable indicator of plant nutrition, pollutant adsorption, and the prevention of leaching from the soil. Given the significance of this property, its measurement in soil studies can be costly and time-consuming, making it one of the challenging, labor-intensive, and expensive tasks faced by researchers. However, previous studies have shown that geostatistical methods can effectively predict and map soil properties. Therefore, this study aims to compare the effectiveness of auxiliary variables and geostatistical methods in producing a map of soil cation exchange capacity in the Sistan Plain.

Materials and Methods: For this research, data from 422 soil profiles in the Sistan Plain were utilized. Sampling locations were selected using a controlled random sampling method. Additionally, this study employed time series remote sensing data, including Landsat 8 OLI sensor images (single bands and indices derived from band ratios), to prepare relevant indices. For this research, data from 422 soil profiles in the Sistan Plain were utilized. Sampling locations were selected using a controlled random sampling method. Additionally, this study employed time series remote sensing data, including Landsat 8 OLI sensor images (single bands and indices derived from band ratios), to prepare relevant indices. These indices were calculated and extracted using ArcGIS software version 10.8.2. Environmental variables derived from the region's digital elevation model were obtained with SAGAGIS software. The corresponding values of the indices extracted from satellite images and the digital elevation model were retrieved at each sampling point in ArcGIS and compiled into tables stored in Excel. A total of 81 environmental variables were examined, including categorized into soil variables, remote sensing variables, and those derived from the digital elevation model. Variables showing significant correlation with soil cation exchange capacity were selected for modeling and geostatistical analyses. Inverse distance weighting methods (with power parameters of one, two, and three), simple and ordinary kriging, and simple and ordinary cokriging were employed as geostatistical approaches for interpolation and prediction of cation exchange capacity values at unsampled locations. The influence of each selected environmental variable was individually assessed within the cokriging methods concerning their effect on the prediction accuracy of cation exchange capacity. To evaluate the accuracy of the various models, two statistics were used: root mean square error (RMSE) and mean error (ME).

Results: The results indicated that the best variogram model for predicting the spatial variability of cation exchange capacity in the soils of the Sistan Plain, based on the minimum residual sum of squares (RSS) and the maximum R² value, is the Gaussian model. The spatial suitability index, calculated as the ratio of local variance to the threshold, indicated that the cation exchange capacity of the soils in the Sistan Plain falls within the moderate spatial suitability class. Correlation analysis revealed that topographic variables derived from the digital elevation model had the strongest correlation with soil cation exchange capacity. An examination of the spatial distribution pattern of cation exchange capacity in the region showed that the lowest values (less than 6 cmol/kg) were found in the southeastern and southern areas. Low to moderate values (6–12 cmol/kg) were observed in the southern, eastern, and parts of the northwestern regions; moderate to high values (12–24 cmol/kg) were found in the central, northeastern, and western areas; and the highest values (greater than 24 cmol/kg) appeared in small sections of the northeast and west. Among the methods tested, ordinary cokriging performed best. When combined with ordinary cokriging, the soil clay percentage among soil variables, NIR_TOA among remote sensing variables, and elevation among variables extracted from the digital elevation model demonstrated the best performance and were selected as the most accurate method for mapping soil cation exchange capacity in the Sistan Plain.

Conclusion: Soil CEC is a key characteristic for managing soil fertility and plant production, making the spatial modeling of this parameter crucial for farmers and land managers. This study found that the average CEC in the Sistan Plain ranged from low to moderate values (12.67 centimoles per kilogram of soil). The spatial correlation of CEC in the Sistan region was moderate, likely due to the complex geomorphology and variations in depositional environments. Geostatistical methods, particularly cokriging with auxiliary variables, can effectively produce geostatistical maps, greatly assisting various land users in better management. Ultimately, ordinary cokriging, which utilized a combination of three selected auxiliary variables, was identified as the best and most accurate method for mapping soil cation exchange capacity in the Sistan Plain. The lowest CEC values were observed in the southeastern and southern parts of the area, while the highest values were found in small sections of the northeast and west. Additionally, CEC values ranged from low to high in the central, northeastern, and western parts of the region. The spatial distribution pattern of soil CEC appears to be strongly influenced by the distribution of soil particle sizes in the area. The results demonstrate that incorporating combined environmental variables that reflect soil formation conditions or influencing factors, alongside cokriging, can lead to accurate and practical mapping outcomes.