A fuzzy logic approach for assessment and mapping of groundwater irrigation quality: a case study of Marvdasht aquifer, Iran

Two shortcomings of the non-numeric United States Soil Laboratory (USSL) water quality designation process are, first, the uncertainty involved with assigning samples to the bordering discrete classes is not considered and, secondly, it is difficult to map. One solution to these may lie in the use of a Mamdani fuzzy inference system (MFIS). The main aims of this study were (1) to compare the MFIS and USSL approach to classifying groundwater quality for irrigation and (2) to explore the spatial variability of groundwater quality for irrigation in the Marvdasht aquifer using the MFIS output. For this purpose, 49 agricultural wells were sampled and their sodium adsorption ratio and electrical conductivity were determined. In 81% of cases, the MFIS led to the same class as USSL and in 92.8% of cases agreed with USSL classification. In only 2 out of 49 samples did the MFIS and USSL strongly disagree. The comparison showed that the MFIS method is more acceptable, reliable, and logical in the classification of water quality for irrigation purposes than other methods. The output map of the MFIS scores indicates that the upper part of Marvdasht aquifer has better quality water for irrigation.