Simulation of groundwater salinity using Artificial Neural Network (ANN) , Particle Swarm Optimization (PSO) and SEAWAT model. (Case study: Debal khazaie sugarcane plantation)

Background and Objectives: Soil salinity is main factors which adversely affect the sugarcane yield in the southwest of Iran. Therefore, assessment and monitoring of these factors are important issue in this area. Due to the large area of sugarcane fields in this area, monitoring of these factors are very time-consuming and costly. computers models can be considered as an appropriate approach for dealing with this problem. Therefore, this research was conducted to find a suitable model for simulation soil salinity in sugarcane fields by using Nural Network models and SEAWAT model. In recent years the use of intelligent models to predict the groundwater salinity is increasing rapidly due to the ease of use and accuracy of these models in the non-linear equations, and complex mathematical returns Taqrib. Use of models saves time and costs and also provides accurate results. ANN methodology has been applied in almost all branches of science with good results during the last decades. (Saey et al, 2009), neural network model (Artificial Neural Network, Particle swarm optimization( to predict Soil used salinity and good performance of the model to predict soil salinity confirmed.


Materials and Methods: In this study, Artificial Neural Networks (ANN), Particle Swarm Optimization) PSO + ANN) and SEAWAT model is used to predict groundwater salinity For this purpose, field R9-11 of the Debal Khazaei sugarcane plantation is selected and number piezometers were installed in different depth and distance from collector. piezometers were in 7 categories, each category includes depths of 2.2, 3, 4 and 5 meters above ground level , was installed in different layers of soil. The volume of irrigation water, salinity of irrigation water and salinity drainage water in this period measurements from November 2013 to October 2014 on a daily basis. Of the problems that exist in the use of Artificial neural networks, the problem is education. In this study, using education PSO ) Particle swarm optimization) method is trying to fix this.

Results: The results showed that the the Particle Swarm Optimization method has a highest accuracy in predicting groundwater salinity. So that the average RMSE in different depths between measured and predicted with artificial neural network, Particle swarm optimization and SEAWAT obtained 0.092, 0.017 and 0.745 , respectively.

Conclusion: Overall, the results of this study showed high accuracy of studied models(Artificial Neural Network, Particle swarm optimization and SEAWAT) for simulation of groundwater salinity that's because accurate measurement of input parameters.

Keywords: Salinity, SEAWAT, Simulation, Neural Network,Mathlab