Optimization of DRASTIC and SINTACS Models According to Geographical Information System with the Use in Analytical Hierarchy Process (AHP) (Case Study: Andimeshk Plain)

Introduction
Because of expansion of agricultural activities and excessive use of chemical fertilizers and location of municipal and industrial wastewater of Andimeshk, it is possible for this aquifer to be polluted. Therefore, the aim of present study is to assess the aquifer vulnerability of Andimeshk plain and recognition of sensitive areas against pollution which this work can be conducted by using DRASTIC and SINTACS models. Although these models are among the most used models for assessment of aquifer vulnerability, therefore these models should be corrected on the basis of local hydrological conditions to obtain correct results. Therefore the main aim of this study is correction of DRASTIC and SINTACS models according to local hydrological conditions and finally determining the best aquifer vulnerable model for the plain.
Materials and methods
Study area
Andimeshk plain is located in the north west of Khuzestan province with the area of 295 km² and its geographical coordinates is in north latitude, 32° 15′  to 32° 35′ and east longitude 48° 29′ to 48° XX′.
DRASTIC and SINTACS models
Generally, parameters used in these models are consist of depth to watertable, net recharge (effective infiltration), aquifer media (aquifer hydrogeologic characteristics), soil media, topography (slope), impact of vadose zone (unsaturated condition) and hydraulic conductivity.
Analytical Hierarchy Process (AHP)
The AHP method is based upon the construction of some series of Pair-Wise Comparison Matrices (PCMs), which compare all the criteria to one other. Saaty (1980) suggests a scale of 1-9 for PCM elements, wherein the value of 1 suggests that the criteria are equally important and the value of 9 shows that the priority of this one is extremely more than the other one in this comparison.
Data acquisition
Based on the collected data and required information, a database in the GIS environment was prepared for calculation of final vulnerability index in the methods. So ArcGIS9.3 software has been used. The main aim of the criteria maps preparation is its modulation by using overlaying technique and finally after that preparation of aquifer vulnerability maps by the models.  Finally to identify the accuracy of specific weights for parameters the models have been analyzed according to their sensitivity to the real conditions. In the present research in addition to the entered parameters according to the DRASTIC and SINTACS model theories the corrected weights by AHP method for preparation of maps indices were used. To determine the optimized weight of each parameter, correlation coefficient calculated vulnerability indices by the methods and nitrate concentration have been calculated by using Simple Linear Regression Analysis (SLRA). Finally according to more correlation coefficient of each index with nitrate concentration, the optimized model has been determined for assessment of aquifer vulnerability. It is necessary to mention that for reduction of decision making errors and correct choice of the weight many expert views were used. Therefore many models were executed for calculation of aquifer vulnerability index.
Sensitivity analysis
In the present research single parameter analysis method has been used to analyze the sensitivity of used models.
Specific vulnerability to nitrate pollution
The Composite DRASTIC index (CD index) is an adaptation of the DRASTIC index based on the addition of a new parameter defining the potential risk associated with land use (L). The risk potential map related to land use using the same methodology applied for other parameters of DRASTIC index will be prepared. It is necessary to mention that for calculation of specific vulnerability of SINTACS method perform according to the same methodology used in the Composite DRASTIC index calculation (Composite SINTACS index or CS index). To prepare specific aquifer vulnerability map of the plain using DRASTICS and SINTACS models at the first step because of the importance of parameters of this area in transfer of pollution into the aquifer, these parameters with the weight of 5 will be added to both the parameters in a normal way. For correction of the weight given to this parameter in the real condition using Simple Linear Regression Analysis (SLRA), correlation of specific vulnerability indices has been determined by using both aquifer models and Nitrate concentration. In this type of vulnerability also the method of weighting based on AHP method has been used for determination of optimized weight according to the real condition of the area.
Results and discussion
Results obtained from overlaying of criteria maps of DRASTIC and SINTACS models show that the final intrinsic vulnerability index for DRASTIC model is between 73 to 157 and for SINTACS model is between 90 to 171. To check the accuracy of specific weight of the both model parameters, single parameter sensitivity analysis has been done based on local condition. Statistical results obtained from the single parameter sensitivity analysis of the models show that the most effective parameter in calculation of intrinsic vulnerability of the studied aquifer unsaturated area parameter (with the average effective weight of 28.59 and 29.32 percent respectively), and topography (with average effective weight of 8.83 and 7.04 percent respectively) have the minimum effect in calculation of this index. Comparison of effective and theoretical weights for DRASTIC and SINTACS parameters models are not fully matched with each other. Considering this inconsistent, parameter weights for both of models have been corrected by AHP method according to hydrological conditions. Results obtained from this method show that among the models executed for DRASTIC model, the best correlation coefficient is 0.752, and it has been chosen as optimized model, because it has better correlation coefficient compared to normal DRASTIC (0.741), also, after determining correlation coefficient in different corrected SINTACS models the greatest correlation coefficient was 0.659 which due to having greater correlation coefficient against normal SINTACS model (0.636), it has been determined as the optimized model. Finally between two determined optimized model corrected DRASTIC model has been determined which is because of having more correlation with the Nitrate concentration as the optimized model for assessment of final aquifer intrinsic vulnerability of Andimeshk plain. Results obtained from Simple Linear Regression Analysis (SLRA) between CD and CS indices with aquifer Nitrate concentration show that the correlation coefficients for both of the indices are 0.651 and 0.529 respectively. After execution of different models by AHP weighting method and determination of their correlation coefficients with Nitrate concentration, the best model of specific aquifer vulnerability model for the plain have been determined according to the most correlation coefficient for both of the methods. specific vulnerability map obtained from corrected CD index has been used for evaluation of possibility for studied aquifer, because of its greater correlation coefficient (0.749 and 0.657 respectively for CD and CS indices), Generally, optimized weight of composite DRASTIC and DRASTIC model according to hydrological condition of Andimeshk plain has been presented in table No.1.
Table No. 1: Optimized weights of DRASTIC and Composite DRASTIC model parameters





Parameter


Primary weight


Optimized weight




DRASTIC


Composite DRASTIC




Depth to watertable


5


3.86


3.93




Net recharge


4


2.03


2.28




Aquifer media


3


1.22


1.35




Soil media


2


2.55


2.84




Topography


1


1


1




Impact of vadose zone


5


5


5




Hydraulic conductivity


3


1.59


1.72




Land use


5


─


1.1




Inconsistency Rate


─


0.035


0.052





Conclusions
Increasing the correlation coefficient of optimized DRASTIC, SINTACS indices, corrected CD and CS indices by AHP method with Nitrate concentration of studied aquifer show that optimization of vulnerability models according to the hydrological condition of the area is more compatible with the aquifer real conditions. Vulnerability index of corrected DRASTIC models is between 51.81 and 122.69.  Also according to obtained results from corrected CD index, specific vulnerability index for studied aquifer is between 65.83 and 135.85. Although nitrate pollution’s risk for Andimeshk plain is not evaluated as high value, due to this area’s condition, urbanization growth, agricultural development, increasing disposal of municipal, industrial and agricultural wastewater, a crucial environmental management should be conducted in this area. This is also possible by cooperation of society members, experts and authorities and correcting of land use planning.