Premium
MODELLING DRAINAGE WATER SALINITY FOR AGRICULTURAL LANDS UNDER LEACHING USING ARTIFICIAL NEURAL NETWORKS
Author(s) -
KAshefipour S.M.,
Sadr M.K.,
Naseri A.A.
Publication year - 2012
Publication title -
irrigation and drainage
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.421
H-Index - 38
eISSN - 1531-0361
pISSN - 1531-0353
DOI - 10.1002/ird.634
Subject(s) - leaching (pedology) , drainage , salinity , irrigation , environmental science , artificial neural network , hydrology (agriculture) , precipitation , water quality , environmental engineering , agriculture , soil science , engineering , meteorology , geology , geotechnical engineering , geography , computer science , soil water , agronomy , machine learning , archaeology , ecology , oceanography , biology
ABSTRACT In this study, an artificial neural network (ANN) method was applied to model the electrical conductivity (EC) of drainage water, based on the effects of irrigation water quality parameters such as HCO 3 ‐ , Cl ‐ , SO 4 2‐ , Ca 2+ , Ma 2+ , Na + , area information (leaching area and planted area), weather conditions (temperature, evaporation and precipitation) and time (days and years of system in operation). Data (312 sets) for the years 2001–2008 were available from a site located in Shoeibieh, south‐west Iran. About 70% of these data were applied to train the model, 20% for verification and about 10% were used for a final test of the ANN model produced. Many runs were carried out with different combinations of the parameters. Comparison of the results obtained for the training, validation and test patterns with the corresponding measured values showed relatively good agreement between both sets of EC values. It was also found that the most important parameters for EC simulation were the years after leaching started and the water salinity of the Karoon River, the main water supply for these lands. Copyright © 2011 John Wiley & Sons, Ltd.