Exploratory analysis of Moscow groundwater composition

The protection and sustainable use of groundwater resources require continuous monitoring and forecasting of both natural and anthropogenic pollutants accumulated in them. However, due to a lack of analytical data, a reliable assessment of these and other trends in the development of indicators necessary for making appropriate management decisions remains a great challenge. Even in a megalopolis such as Moscow, time series of data from available observation wells are limited to 10–20 members, which underscores the inefficiency of traditional statistical research methods. To solve the above problem, this article proposes an expert statistical exploratory data analysis (EDA) approach based on graphical visualization of statistical results and, compared to traditional methods, less dependent on the amount of experimental data. Using data obtained from the observation wells of the Teplostanskaya Upland in the Southwest Administrative District of Moscow during 2018–2023, statistical heterogeneity was established in the time series of groundwater composition and properties, eliminating randomness in the formation of data and indicating the presence of hidden patterns in their change, mutually consistent variability was detected among certain pairs of groundwater pollutants, the possibility of identifying the origin (natural or anthropogenic) of individual groundwater components was demonstrated, and the feasibility of reducing the number of variables in a set of controlled indicators through the application of principal component analysis (PCA), accounting for 72 % of the variance in the problem under consideration, was shown. Since all measurements were carried out using a set of spatially separated wells, the experimental data were studied by common panel analysis method.