Groundwater vulnerability model optimization using GRG2 algorithm

Industrialization and human overpopulation are among the major reasons of environmental pollution and especially the pollution of water resources systems. Human activities can contribute to upsetting the delicate balance of nature by polluting air, soil, and water resources. The environment has the ability to recover from damages caused by pollutants, however, the problem is that this ability is limited and the process is time consuming. Nowadays, studying groundwater systems and providing new methods and managerial tools has become a widespread activity among water resources planners and managers. Since majority of Iran’s water demand is provided by groundwater and aquifers, there has to be a meaningful accentuation and notes on both qualitative and quantitative characteristics of water extracted from this kind of water resources systems as a strategic and a decisive phenomenon in regional planning. Groundwater, which is a major source of domestic, industrial, and agricultural water supply worldwide, is less susceptible to different types of water pollutants in comparison with surface water. Nevertheless, monitoring and remediation of groundwater pollution are more complicated and expensive than that of surface water pollution. Spatial variability and lack of data make groundwater remediation more difficult or even impractical. Therefore, pollution prevention must be considered as an immediate action for an effective management of groundwater resources.
The vulnerability assessment of groundwater system is a good practice for modelling and planning to preserve and protect these valuable systems from pollutions. Quantifying how and where groundwater systems are prone to be polluted is termed as vulnerability assessment. The tendency of contaminants originated from the land surface to reach water table is termed as groundwater vulnerability. National Research Council defines two types of groundwater vulnerability: specific vulnerability and intrinsic vulnerability. Intrinsic vulnerability is independent of the nature of water resources system, while specific vulnerability takes into account the susceptibility of groundwater to particular contaminants. Groundwater vulnerability assessment is a tool for protecting and qualitative management of these valuable resources. So, in this paper a well-known groundwater vulnerability model is used to assess aquifer vulnerability to pollutants. Different approaches such as indexing, statistical analysis, and process-based methods have been proposed for assessing the vulnerability of groundwater to pollutants and delineating vulnerable areas of aquifers. GOD, DRASTIC, AVI, SINTACS, RISKE, and DRAV are examples of indexing methods. This study proposes Dehgolan aquifer vulnerability assessment using DRASTIC technique. DRASTIC index is calculated as a linear combination of factors affecting the groundwater vulnerability to pollutants. In the present study, Generalized Reduced Gradient 2 (GRG2) algorithm was used to create an optimized weighted linear combination of the factors used in DRASTIC method based on gridded images to establish the aggregated vulnerability model. As a good criterion to evaluate the efficiency of this model, Spearman Correlation Rank Test (SCRT), between Groundwater vulnerability Index and Nitrate (NO3) concentration was carried out that was equal to 0.493. The optimization-simulation in this study includes a simulator of DRASTIC model with a given vector of candidate weights by optimization algorithm and the GRG2. The GRG2 algorithm could reach the desired value of SCRT with the minimum number of iteration. Then for evaluating this criterion, model’s parameters were optimized using (GRG2) algorithm and the correlation coefficient as objective function.
The optimized correlation was computed as 0.651. The study results, found optimization procedure as good idea in vulnerability assessment, where there is low or insignificant relationships between what is modeled using vulnerability technique and extend of observed pollutions in groundwater system. Optimization process showed the approach taken in the study useful in balancing and deriving reasoning multiplicative weights in a realistic groundwater vulnerability assessment. Even though, the study proposes a procedure for weights delineation of DRASTIC vulnerability method, it does not optimize number of subdivisions and sub-weights. Hence, further study would tune the method so that describes vulnerability condition concordant to observed pollutions. This study showed that although in vulnerability assessment, inaccurate data and uncertainty caused poor relationship for the vulnerability index and the current condition of the aquifer, the optimization approach could tune a desirable groundwater vulnerability map for regional planning. Finally, it is to say, the procedure which is described in this paper can be used for many aquifers and different groundwater systems for assessing their accurate vulnerability condition.