PREDICTION OF NITRATE CONCENTRATION IN GROUNDWATER USING MLR AND AGRICULTURAL INDICATORS

Environmental pollution results in the degradation of all segments of nature. Groundwater, as one of thesignificant sources of drinking water at the global level, is increasingly exposed to various types of pollutants thatare mainly the result of anthropogenic action. Nitrate are one of the pollutants that can persist for a long time ingroundwater and their concentrations can reach high levels. Namely, the fact that groundwater is used for variouspurposes such as: irrigation, food production, drinking, special attention is focused on monitoring the concentrationof nitrate in groundwater. In this sense, EU legislation also prescribed a standard for drinking water in terms ofnitrate concentration in order to avoid possible threats to human health. The main sources of nitrate in groundwaterare agricultural activities, which primarily involve the use of chemical fertilizer, animal farms, waste water fromindustries that produce chemical fertilizer. In this paper, the prediction of average annual concentrations of nitrate ingroundwater at the national level was performed. Bearing in mind the mentioned sources of nitrate in groundwater, atotal of seven agricultural indicators were selected in this paper, which were assumed to contribute to higher nitrateconcentrations, namely: the area used for agriculture, meat production on farms, the balance of indicators - whichrepresents the total potential environmental risk caused by excess or deficit of N and P in agricultural soils, areaused for organic agricultural production, consumption of chemical fertilizer, sale of pesticides and percentage ofpopulation associated with at least secondary wastewater treatment. The prediction was made using multiple linearregression, where agricultural indicators were used as input parameters, i.e. independent variables, a total of seven,while the dependent variable was the average annual concentration of nitrate in groundwater at the national level. Inthis paper, five EU countries were selected for nitrate prediction: Belgium, Bulgaria, the Czech Republic, Germanyand France. For the development of the MLR (Multiple linear regression) model, available data from 2011 to 2015were used for all eight variables. Before developing the model, a correlation analysis was first performed todetermine whether the selected independent variables were correlated with the dependent variable. The results of thecorrelation analysis showed that there is a significant correlation between the independent