Basirat, Mohammad (2019) Predicting Exchangeable Sodium Percentage (ESP) by Artificial Neural Network Using Readily Obtainable Soil Properties in the Qorqori Area, HeIArmand County. Masters thesis, University of Zabol.
|
Text
Predicting Exchangeable Sodium.pdf Download (185kB) | Preview |
Abstract
Considering the great extent of salinity and sodium soils in Iran and the fact that its agricultural lands have constantly been salty followed by shortage of land resources suitable for agriculture, various methods are required to improve the soils. Today, the use of artificial neural network methods is very common in predicting soil properties. ESP and SAR are two components indicating salinity and sodium contents of soils. The present study was carried out to evaluate the properties of sodium absorption ratio and exchangeable sodium percentage using artificial neural network and regression method in Qorqori region, Hirmand, Sistan and Baluchestan. For this purpose, 130 soil samples were collected from the region as regular grids for thorough evaluation of their physicochemical properties. To predict these two properties using the MLP neural network and the regression method, EC and PH were used which had the highest correlation with ESP and SAR. Finally, the best network was selected by R2 and RMSE. For ESP, the MLP neural network with 10 neurons in the hidden layer, the maximum value of R2 = 0+13, and the root mean square error of RMSE= 0+0245 provided a better estimate. For SAR, the MLP neural network with 1 neurons in the hidden layer, the maximum value of R2 = 0+11 and the lowest root mean square error of 0+0117 = 0 RMSE, offered a better estimate. The results of regression analysis and artificial neural network showed that the regression method with R2 = 0+70 and RMSE = 1+75 had a weaker performance than the neural network and only 707 of the variations in SAR and ESP were justified in the region. Since measuring SAR and ESP features has a high cost and is time- consuming, the artificial neural networks can be used to identification, determination, prediction, and modeling of the sites sensitive to salinity and sodium.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | ESP, SAR, Artificial Neural Network, Qorqori region |
| Subjects: | S Agriculture > S Agriculture (General) |
| Depositing User: | Mrs najmeh khajeh |
| Date Deposited: | 26 Jun 2019 04:13 |
| Last Modified: | 26 Jun 2019 04:13 |
| URI: | http://eprints.uoz.ac.ir/id/eprint/2494 |
Actions (login required)
 |
View Item |
