Groundwater level measurement and monitoring are one of the basic and essential steps in groundwater studies. Data in geographical studies, especially in groundwater studies, are generally measured at points in monitoring stations and play an important role in analyzing temporal and spatial variations of phenomena. optimization is one of the main issues in reducing the cost and increasing the quality of the extracted data. The main purpose of this research is to optimize the groundwater monitoring network of the Arak plain aquifer, in which a new metaheuristic method called Colliding Bodies Optimization (CBO) is used. For this purpose, the groundwater wells data are collected and refined. Then the CBO method is implemented and the inverse distance weighting (IDW) method is used to calculate the cost function in CBO. In order to evaluate the accuracy of the output, the results were compared with the ant colony optimization (ACO) method. Arak plain is located in the watershed of Miqan desert wetland and its groundwater level has been decreasing in recent years. Continuous and accurate monitoring of groundwater levels in Markazi province and especially the Arak aquifer is one of its main needs.
For this purpose, the groundwater wells data are collected and refined. Then the CBO method is implemented and the inverse distance weighting (IDW) method is used to calculate the cost function in CBO. The CBO algorithm is based on simulating the search space with an environment in which the kinetic energy and momentum of the colliding particles are decreasing. In the proposed approach, using the IDW method, a continuous surface was created from the selected stations, and the error generated in the unselected stations was calculated based on the Root Mean Square Error (RMSE) formula. IDW is one of the simplest spatial interpolation methods that has been used extensively in network optimization.
In this study, the average annual data of 57 groundwater level monitoring stations in 1397 were used. Out of 57 monitoring stations, only 43 were active and in 14 stations, measurements were not recorded. Evaluation of outlier data based on Grubbs test showed that well data No. 15 was as outlier data which was excluded from the calculations. In this study, different scenarios were evaluated for removal of 1 to 12 monitoring wells and the curve of the number of removed wells against the amount of error created in the wells was drawn. It was obvious that the error value would increase as the number of deleted stations increased. Comparison of the optimized error percentage shows that the CBO method always has a lower error than the ACO. In general, based on the location of the unselected stations in each scenario, it was observed that the study area is divided into three general parts, north, south, and southeast. In the first three scenarios, the unselected wells are located in the northern part of the aquifer. From scenario four, station No. 6 from the southeastern part is always removed, and from scenario five, station No. 40 from the southern part is always removed as a priority. According to the error diagram, the location of the unselected wells in different scenarios, and also expert opinions, it was found that by removing 6 wells (wells with numbers 5, 6, 9, 30, 36, and 40) from the groundwater monitoring network of Arak aquifer, a maximum of 35 cm of accuracy will be reduced in the well. On the other hand, it saves money and time for data collection. The location of the unselected wells shows that the wells around Miqan Wetland are of great importance in estimating the groundwater level of the aquifer and most of the removed wells are located in the northern and central part of the aquifer. This study also showed that the object collision optimization method is a suitable method in optimizing groundwater monitoring networks.