عنوان مقاله [English]
The reduced groundwater levels of plains have led to increased water extraction cost, increased energy consumption, reduced water quality, and the appearance of subsidence. Structural subsidence in plains could be directly resulted from reduced level of groundwater as well as destruction of alluvial texture in aquifers. Although it results from the compacted underlying layers of soil, subsidence represents unpleasant outcomes in the future. The compacted underlying layers and reduced water table in groundwater basins reduce the water storage space. In other words, reduced groundwater storage is a rational consequence of subsidence (Sharifikia et al., 2014). Scientific investigation and various instruments, including mathematical models, are required to overcome the problems of groundwater resources (Gaura et al., 2011; Hu et al., 2010). Mathematical models allow for investigating changes in the current and future situations of groundwater tables by incorporating different influential factors (Yaoutia et al., 2008; Zhang, 2010). Therefore, using mathematical models, the situation of an aquifer can be simulated by collecting information on the inputs and outputs of the groundwater system at a small cost in a short time. Such simulation models can provide the mutual effects of surface water and groundwater in short- and long-term periods. MODFLOW is among the most important mathematical models (Lachaal et al., 2012).
In this current paper the approaches to minimize subsidence in Abhar plain is evaluated that based on an empirically derived relationship between cumulative subsidence rates and groundwater levels is used with common groundwater model software MODFLOW. Also, the vertical changes in the ground structure were nonlinearly modeled based on the cellular changes of the finite difference method. To determine the essential factor of the changes, the fuzzy model approach and spatial-statistical analysis were adopted. The largest effect on the appearance of subsidence was studied by the regression relationships of the corresponding points in the spreadsheet setting.
-Material and methods
Abhar Plain is located in the northwest of the Namak Lake Basin. It occupies an area of 1926.5 km2, 1040.92 km2 of which is plain, while the remaining area is composed of mountains. The maximum and minimum elevations of Abhar Plain are 2166 and 749 m, respectively. The saturated area of the plain is approximately 657.9 km2 – the corrected area is 657.4 km2. The groundwater extraction area of the plain includes 1359 water wells with an annual discharge of 233.36 million m3, 169 fountains with an annual discharge of 4.88 million m3, and 101 aqueducts with an annual discharge of 1.3 million m3.
Considering that parameters involved in the calculation of the final subsidence layer had reduction or enhancement effects on each other and given the descriptions on the overlapping functions, the current study employed the gamma operator. Computation was performed by the MODFLOW-2005 engine in GMS v.10. The study period was selected to be 119 months, which could be made more accurate and rebuilt based on the maximum available data. The computation engine of PCG2 with 100 outer and inner iterations, a critical convergence variation limit of 0.01 m, and a critical convergence error limit of 0.01 m3 per day was selected. 75% of the interval length was used for calibration in non-stable conditions. After seven executions of the calibration model with a certain number of internal iterations, the optimal final number of surface feeding, horizontal hydraulic conductivity and horizontal hydraulic conductivity anisotropy, specific yield in the form of pilot extraction points, and transferability parameters in the boundaries, and the waterway network in the form of the bunches of lines. The layers were also used to develop the fuzzy subsidence model.
According to the described fuzzy theory, each of the basic parameters influencing or representing subsidence was transformed into a standard raster map in the range of 0-1 by a linear equation. The fuzzified layers included water level reduction, the difference between the initial and final water levels, horizontal hydraulic conductivity anisotropy, horizontal hydraulic conductivity, saturated aquifer depth, surface feeding, extraction flow rate, and specific yield. Direct fuzzification was applied to all the parameters, except for surface feeding, to which inverse fuzzification was applied. Table 2 shows the zoning errors of the selected layers. The parameters of Table 2 were used to develop the fuzzy model. The coefficients of the aquifer were merely extracted from the last calibration round of the groundwater flow model.
In the next step, to investigate the spatial variations of subsidence occurrence, the land-use layer was utilized as the statistical analysis basis of the subsidence raster output.
The results of the current study indicated that subsidence modeling can be carried out by linear regression analysis with a determination correlation coefficient of 70%. Among the influential parameters, merely the spotted aquifer depth layer had a correlation of 30%. The fuzzified variations of the layers with the enhancement-reduction gamma combination had the highest correlations with the spotted subsidence layer of the MODFLOW output. However, the correlation was noticeably lower than expected. In the subsidence model, the final raster layer was transferred to the GIS setting using dispersed points. After zoning using the regional analysis command in two manners, the separate indicators of land-uses and a basic land-use set were extracted. These results revealed that the urban land-use accounted for 26% of subsidence, even though it occupied merely 4% of the aquifer’s area. The agricultural land-use (including gardens), which accounted for 42% of the aquifer’s area, involved 56% of subsidence events. Finally, occupying 54% of the aquifer’s area, the idle land-use accounted for merely 19% of the vertical changes of the aquifer’s structure. The high subsidence in urban areas was noticeable. Considering the alluvial groundwater yield, the effect of water level reduction can be seen with small spatial distances.