The smoothing methods used in assessing the influence of pollution sources on groundwater quality - a case study of metallurgical landfill in Lipówka (southern Poland)

Abstract

Several methods have been used to determine trends in hydrogeochemical elements. This study is concerned with the ordinary 3-period moving averages, the 3-period moving median and the exponential smoothing method. It has taken into account the chloride concentration in groundwater from the region of the landfill of the Katowice Ironworks (southern Poland). The data from two piezometers were used: T1 screened in the Triassic and Q1 screened in the Quaternary aquifers. The main aim of this article was to compare statistical methods and to choose the most appropriate method for an assessment of the impact of pollution sources on groundwater quality and determining any trends. The choice of chlorides as the analyzed indicator is connected with their chemical conservative character. They are also indicators of negative impact on groundwater of e.g. municipal, metallurgical landfill sites on groundwater. Results showed that the moving median is less sensitive to outliers than the moving average. The running median preserves sharp discontinuities in the signal but the biggest smoothing was observed in the case of exponential smoothing. All of these methods filter out the noise and transform the data into a smooth curve, which is unbiased by outliers. Application of the smoothing method allowed the generalization of the monitoring data which clearly showed trends. In the absence of reliable data for short periods, these statistical methods made it possible to fill in missing values. These methods are easier to calculate than regression models even if they ignore complex relationships in the data. Moreover, they can also be used as a component in many other indicators such as the Moving Average Convergence Divergence.