Abstract:
Cation exchange capacity (CEC) is one of the most important soil attributes which control some basic properties of soil such as acidity, water and nutrient retaining capacity. However, the measurement of cation exchange capacity in large areas is time consuming and requires high expenses. One way to save time and expenses is to use simple soil covariates and geostatistical methods in mapping CEC. Therefore, the aim of the present research was to investigate the role of soil covariates in the improvement of spatial variability of CEC. The study area is located in southwest Iran on the Aghili plain, Gotvand, Khuzestan province. In this study, ordinary kriging and cokriging methods were used to predict CEC. 107 soil samples were gathered on a random grid of 200-700 m. 74 samples were used for training and 33 samples for testing the results. A principle component analysis was performed for covariate selection. Clay was selected as a covariate in cokriging due to high correlation between clay and CEC in the first principle component analysis. Based on the cross validation result of predicted dataset, RMSE and ME for cokriging were 2.16 and 0.03 cmol (+)/kg respectively, and 3.36 and 0.09 cmol (+)/kg for kriging, respectively. Based on these results, cokriging performed better than kriging for predition of cation exchange capacity since it used a covariate such as clay, for the improvement of CEC spatial prediction.
Machine summary:
Based on these results, cokriging performed better than kriging for predition of cation exchange capacity since it used a covariate such as clay, for the improvement of CEC spatial prediction.
The main purpose of the current study was to compare soil CEC predictions by kriging and cokriging using the principal components.
With the aim of using the physical-chemical properties of soil’s PC1, as the auxiliary variable for cokriging of soil CEC, PCA was performed on the prediction dataset.
3. Evaluation criteria For testing and the predictions, the performance of kriging and cokriging methods were evaluated using mean error (ME) and root mean square error (RMSE) between the measured soil CEC of 33 soil samples.
/ Desert 24-1 (2019) 99-108 105 The prediction map of soil CEC by cokriging and kriging methods is shown in Figure 6.
Results of validation and cross-validation of kriging and cokriging methods for soil cation exchange capacity in Gotvand, Khuzestan, Iran Variable Min Max Mean Std. dev.
The predicted soil CEC for the data test-set by kriging ranged from 8.
This showed that organic carbon had a moderate correlation with CEC and can be used as a second auxiliary variable in cokriging to improve the accuracy of soil CEC prediction.
This showed that organic carbon had a moderate correlation with CEC and can be used as a second auxiliary variable in cokriging to improve the accuracy of soil CEC prediction.
The predicted soil CEC for the data test-set based on kriging and cokriging methods had some differences in the measured values.