Comparison of Soft Computing and Regression Techniques to Calibrate Electromagnetic Induction in Ardakan

Document Type : Research Paper


1 Assistant Professor, National Salinity Research Center, Yazd, Iran

2 Assistant Professor, Faculty of Agriculture and Natural Resources, University of Ardakan, Iran

3 Professor, Faculty of Agricultural Engineering and Technology, University of Tehran, Karaj, Iran

4 Instructor, National Salinity Research Center, Yazd, Iran


Up to now, different methods have been applied to calibrate electromagnetic induction data. Therefore, at present research, we applied multi-linear regression (MLR) and artificial intelligence techniques (i.e. ANFIS, GA, ANNs) to calibrate the apparent electrical conductivity (ECa)- measured using an electromagnetic induction instrument and electrical conductivity (ECe)- measured in saturation paste. 600 soil samples collected from Ardakan in central Iran and divided into two subsets for calibration (80%) and testing (20%) of the models. To evaluate models, some evaluation parameters such as root mean square, average error and coefficient of determination were used. Results showed that the ANFIS model gives better estimation than the other techniques whereas this model increased accuracy of predictions about 9, 9, 5 and 2% for EC15, EC30, EC60, and EC100, respectability. Higher performance of ANFIS to predict soil salinity might be because of uncertainty. After ANFIS model, GA and ANN had better accuracy than multivariate regression. As a whole, results indicated that artificial intelligence methods had higher performance than regression technique to calibrate electromagnetic induction data.