Assessing the groundwater vulnerability with DRASTIC and Fuzzy-AHP models and validating the results based on the nitrate concentration (Case study: Fumanat area, Guilan province)

ABSTRACT
Groundwater is one of the most important sources of water supply in agriculture and drinking water, nowadays. These sources are very vulnerable to surface pollutant sources such as chemical and animal fertilizers, then identifying areas with high vulnerability is one of the great importance. The aim of this research is to findout vulnerable groundwater areas in the Anzali watershed (Fumanet sub-basin) using DRASTIC and Fuzzy-AHP models. The DRASTIC model uses seven data layers or parameters for modeling, including water depth (D), net recharge (R), aquifer saturation environment (A), soil environment (S), topography (T), impact of vadozone (I), and hydraulic conductivity (C), then it uses fixed weights for input parameters and fixed rankings for sub-parameters. The Fuzzy-AHP model was used to improve the weighting in the DRASTIC model. By validating 20 nitrate wells located in the Fomanat region, using multivariate and univariate linear regression, the output of the Fuzzy-AHP model improved the results compared to DRASTIC. In the vulnerability map produced, the DRASTIC method showed that 0.18% of the area had low vulnerability, 11.22% medium, 58.33% high and 30.42% very high vulnerability. Then, in Fuzzy-AHP, 99.6% low vulnerability, 13.11% medium, 56.45% high and 23.43% very high vulnerability were identified. Both models were successful in identifying areas with medium and high vulnerability risk, and the correlation of the Fuzzy-AHP model with the nitrate map of the region was positive.
Waters are divided into two categories, surface and underground. surface waters are more exposed to pollution, but with a general view, it can be understood that underground waters are exposed to human settlements so the possibility of sewage entering them is higher. Groundwater supplies more than 60% of irrigated agriculture and 85% of drinking water resources. underground aquifers in areas that are more populated and economically rich is decreasing. Groundwater as the most important source of water supply plays an important role in agricultural, drinking and industrial uses. Water scarcity occurs in all populated continents. The increase in population also causes the demand for more food resources and the use of chemical fertilizers and pesticides. Throughout history, groundwater has been inseparable from human life and sustainable agricultural production, but it is not evenly distributed around the world.
In Iran, the best source of drinking water supply is underground water. One of the main human inputs for the physical and chemical pollution of underground water is urban and industrial wastewater, which is increasing with population growth, urbanization and lifestyle changes. Groundwater pollutants include organic and inorganic pollutants such as arsenic, mercury, aluminum, lead, fluoride, nitrate, iron, pesticides, chlorinated solvents, where nitrate from fertilizers and animal waste is the most common pollutant. Vulnerability assessment is an essential part of land use planning and zoning protection approaches for groundwater protection. Identifying high risk areas of contamination is essential for healthy management of groundwater resources. Generally the environment is a phenomenon that does not exist in all parts of the world in the same way and any model in any area may not have the same output according to the altitude of the area (flat, hilly or mountainous) and the type of aquifer. Of course it is clear that each model has its advantages and disadvantages so the better results of one model compared to another model in the vulnerable area do not mean that model is rejected.
The aim of this research is to find the vulnerable areas of underground water in the Anzali watershed (Fumanat sub-basin) using DRASTIC and Fuzzy-AHP models in the geographic information system. The land area of Fumanat includes rice fields, tea gardens and fish breeding ponds. The DRASTIC model uses seven data layers for modeling, including water depth (D), net recharge (R), aquifer saturation environment (A), soil environment (S), topography (T), unsaturated zone influence (I), conductivity and hydraulic (C). This model has a fixed weight for the input parameters and a fixed ranking for the sub-parameters that are below the criterion of the input parameters.
 
Methodology
First, we rank and weight the input layers of the model. The data includes the boundary of the study area, pumping test data, the location of nitrate wells and their concentration, piezometric wells and the water depth of the wells, the layers related to the type of underground soil, the amount of rainfall in the area along with the location of the stations. The elevations of the area and the slope were obtained from the ASTER satellite data with an accuracy of 30 meters and the soil layer of the area was also obtained from the Google Earth Engine system.
The map layers of water depth, aquifer feeding, soil environment, aquifer environment, influence of unsaturated zone, hydraulic conductivity and permeability were prepared using IDW and Kriging methods, depending on which interpolation model had the best adaptation to the area.
 
Results and discussion
Fuzzy-AHP model was used to improve weighting in DRASTIC model. By validating 20 nitrate wells in Fumanat region, using multivariate and univariate linear regression, Fuzzy-AHP model had a better output than DRASTIC.
For the Fuzzy-AHP method, the same layers classified in the DRASTIC method were used as input, but first, the layers were standardized using the Fuzzy membership command, and then they were divided into four regions with the Reclassify command, and finally, with AHP model and ranking, A vulnerability map was generated.
The results of the DRASTIC method showed that 0.18% of the area had low vulnerability, 11.22% had medium vulnerability, 58.33% had high vulnerability and 30.42% had very high vulnerability, while in the Fuzzy-AHP model, 6.99% was identified as low vulnerability, 13.11% moderate, 56.45% high and 23.43% very high. By improving the weighting and using fuzzy functions to standardize the inputs and eliminate the uncertainty in the collected data, the vulnerability map had a better match with the nitrate map of the area.
Nitrate does not occur naturally in the ground and enters the ground through surface contaminants, so it is used as a reliable indicator of groundwater vulnerability. Nitrate ions are usually found and measured in wells located in high-risk areas for groundwater pollution. The average nitrate measurement (seasonally in Fumanat area) was used. The nitrate concentration map of the region was also prepared by interpolation method from 20 wells in Fumanat, and then the value of those wells was obtained in all three vulnerability maps produced. Finally the relationship between them was discussed with regression model.
 
Conclusion
The validation and comparison of the two methods were done. The distribution of nitrate concentration is higher in the west and southwest of Fumanat region, and the concentration of nitrate is higher in the higher altitudes. The results showed that both models were successful in identifying medium and high vulnerable areas, but the correlation of the Fuzzy-AHP model with the nitrate map of the area was positive so this model had better output due to reality of vulnerability.
The high vulnerability was due to the shallow depth of the water table, the unsaturated zone and the high amount of net nutrition under the Fumanat basin. Based on the results of the concentration of nitrate wells, the areas with higher altitude had more pollution, which was somewhat consistent with the DRASTIC model and Fuzzy-AHP. The cause of high pollution in high areas can be high rainfall and low water depth of piezometric wells, which causes a large amount of surface pollution to enter the underground water table.