Estimating soil salinity level by pedotransfer functions

Abstract In this study, the electrical conductivity (EC) of cultivated fields were predicted by pedotransfer functions (PTFs) using basic soil properties as variables in stepwise analyses. Before using PTFs, 207 soil samples were divided into development and validation datasets. This division was done 10 times to assess the accuracy and reliability of the PTFs. According to the principal component analysis (PCA), the EC values had higher load in PC1 and PC2 with clay, exchangeable Na, Ca, Mg and K contents. The EC values had significant positive correlations with pH, clay, exchangeable Ca, Mg, K, Na content, and negative correlations with silt and sand content. According to the stepwise analyses, 10 PTFs or linear multiple regression models were obtained using development data sets. The accuracy in development data sets and reliability in the validation data sets for these PTF models were assessed with R2 and RMSE values. The higher mean R2 and the lower mean RMSE values were obtained in the development data set when compared to the validation data set. The PTF-7 including clay, exchangeable Na, Mg and Ca was suggested to predict soil EC values of cultivated fields. It was determined that clay and exchangeable Na, Mg and Ca content were the most effective soil properties on predicting the EC values of cultivated fields. Keywords: Salt, EC, exchangeable cations, soil texture