Evaluating geostatisical approaches and their integration with pedotransfer functions for estimating the spatial distribution of cation exchange capacity

Abstract

Cation exchange capacity (CEC) is a fundamental soil property that governs nutrient retention and availability, directly impacting plant growth and environmental risk assessments related to heavy metals and organic pollutants. Accurately mapping the spatial distribution of CEC is essential for sustainable soil management. This study aimed to identify the most effective interpolation technique for predicting soil CEC distribution in Hunan Province, China, using optimized data. A total of 467 soil samples were collected from a 0–30 cm depth, and soil texture, organic carbon content, and CEC were analyzed. Four spatial interpolation methods—Kriging, Co-Kriging, Fuzzy Kriging, and Regression Kriging—were compared. Results indicated that Regression Kriging outperformed other methods, achieving a coefficient of determination (R²) of 0.66, a root mean square error (RMSE) of 3.92 cmol +. kg⁻¹, and a mean square error (MSE) of 15.39 cmol +.kg⁻¹. Furthermore, integrating the Adaptive Neuro-Fuzzy Inference System (ANFIS) algorithm for data optimization significantly enhanced prediction accuracy. When Regression Kriging was applied to the ANFIS-optimized dataset, model performance improved substantially, with R² increasing to 0.88, RMSE decreasing to 2.87 cmol +. kg⁻¹, and MSE reducing to 8.28 cmol +.kg⁻¹. These findings underscore the importance of coupling ANFIS-based data selection with advanced spatial interpolation techniques to achieve more precise and reliable predictions of soil CEC distribution, contributing to improved land management and environmental sustainability.