Evaluation of Irrigation Water Quality in The Western Mediterranean Basin Using Fuzzy Logic Methodology

Abstract

Irrigation water quality is largely determined by a graph based on Sodium adsorption rate (SAR) and Electrical Conductivity (EC) developed by the United States Salinity Laboratory (USSL). Limits are flexible as irrigation water quality values cannot be expressed as a net value. As a result of the classifications based on EC and SAR, irrigation water quality is expressed linguistically rather than numerically. In this study, irrigation water quality was evaluated by multiple linear regression (MLR), Mamdani inference system (MFIS), and adaptive neuro-fuzzy inference system (ANFIS) using EC and SAR values measured by the Electrical Power Resources Survey and Development Administration between 2003 and 2008 at 10 stations in the Western Mediterranean Basin of Turkey. The Mamdani method was used in the study and compared with the traditional graphical method. ANFIS model was found to have a much better prediction ability than MLR and Mamdani fuzzy inference system (MFIS).