Sensitivity Analysis of The Urmia Plain Aquifer Level

Introduction
Water is the source of life and a strategic resource for human societies. The need for this vital resource is increasing exponentially due to the increase in population and the development of industry and agriculture. People are forced to use underground water because surface water is not generally and permanently responsive to diverse needs. A decrease in their volume and many problems have been caused by the excessive use of these resources. This crisis has caused regional crises caused by the imbalance of resources and consumption, along with climate changes, has raised the issue of integrated management of water resources more than ever. Agricultural land has been developed due to the increase in population and the need for more food. Programs without principles that rely solely on the quality and quantity of underground water resources have been harmful. Groundwater aquifers are transformed into sources of the country's needs due to the heterogeneous and untimely temporal and spatial distribution of discharges and surface water flows. In recent years, with the increase in water demand and the non-supply of a significant part of it by surface water sources, the extraction - permitted and unauthorized - of underground water sources has been given much attention; so that the level of underground aquifers has decreased dramatically across the country. The purpose of the present study was to investigate the impact of the important variables of precipitation, inflation and annual population as a representative of climatic, economic and social factors on the fluctuations of the underground water level in Urmia region.
 
Material and Methods
In the present study, the impact of three factors, precipitation, population and inflation, on the subsidence of the Urmia Plain aquifer has been investigated. To do this, multiple linear regressions was performed between the data of the annual loss of the groundwater level during 38 years, 1981 to 2019 with three variables of precipitation, population and inflation index of the previous year. According to the previous researches, firstly, an index of inflation has been established by comparing the average loss of the piezometric level of the underground water in


 
Urmia region as a dependent variable, with the three independent variables of the average rainfall of water year as the most important climatic factor, the annual population of the major centers of human concentration located in the Urmia plain of previous year, and the base coefficient of the annual monetary value of previous year compared to 1981 using a multivariable linear regression. Then, the outcome is compared to the outcomes of artificial neural networks such as four-layer perceptron, three-layer perceptron, and radial basis function. All three networks have an input layer with three neurons to receive the values of the three independent variables of precipitation, population and inflation. One or two hidden layers with a number of neurons, to perform calculations and process the relationship between independent and dependent variables; and an output layer with a neuron to provide the processing results i.e., the estimated aquifer subsidence rate. The data used in the present study were derived from the years 1981-2019. The reference of the aquifer level data is the hydrograph extracted from 67 piezometer wells in the area by the underground water unit of basic studies of the West Azerbaijan Regional Water Company. The annual rainfall data reference is of the Urmia camp evaporation station located in the company premises, which is well controlled and highly reliable as an indicator of rainfall changes in the region. Population data is sourced from the Iranian Statistics Center, while inflation data is sourced from the Central Bank of Iran.
 
Results and Discussion
According to the results of the reviewed models, despite the differences in the values of the numerical results, in all four models: multivariate linear regression, perceptron artificial neural networks of the four layers MLP:3-2-2-1, and the three layers MLP:3-5-1 and the radial basis function RBF: 3-5-1, it can be seen that the importance of the independent variables under study are population, inflation and annual precipitation respectively. It is obvious that a larger population needs more food, clothing, housing, etc., which, according to the concept of virtual water, ultimately leads to more use of the limited available water and soil resources. Economic activity, particularly agriculture, is increased due to the depreciation of currency and decrease in people's purchasing power, which is a result of the decrease in purchasing power and the depreciation of currency. This problem has also led to the change of land use of natural resources to agricultural lands that are either rainfed or irrigated. Explaining that rain fed lands cause more rainwater loss through capture and then evaporation and transpiration by plants planted by farmers. Irrigation of agricultural plants or gardens of irrigated lands - mainly with unauthorized water harvesting - ultimately leads to more water consumption. Additionally, humans have exploited underground water resources due to the inappropriate and untimely distribution of rainfall and surface water resources. Although by adopting new management methods, both social and economic, and improving water productivity, despite the increase in demand for water, despite our efforts to protect this vital, sensitive, and strategic resource, statistical studies, including the current results, demonstrate that we have not chosen the correct solutions. Considering some irreparable effects of the aquifer level drop, including irreversible changes in the mechanical characteristics of the soil, which lead to more vulnerability of infrastructures and facilities; the emphasis is placed on comprehensive water resource management and the concept of virtual water and its trade.