Using multidimensional analysis to identify water pollution markers

• L.V. Novikova

  Kazan Federal University, 18, Kremlyovskaya str., 420008, Kazan, Russian Federation

  Автор

• N.Yu. Stepanova

  Kazan Federal University, 18, Kremlyovskaya str., 420008, Kazan, Russian Federation

  Автор

Oil and gas production remain a primary source of pollution for
aquatic ecosystems. However, the lack of specific markers complicates the
identification of parties responsible for cumulative environmental damage. The aim
of this study was to identify a set of indicators specific to oil production activities
through multivariate statistical analysis. The study was conducted using 232
surface water samples collected from the Sheshmа River, with a catchment area
dominated by oil extraction, compared to the Myosha River, where the catchment is
predominantly agricultural. Thirty parameters were analyzed, including standard
water quality indicators, elemental analysis, and organic pollutants (COD, BOD5).
The following methods were used: descriptive statistics (Shapiro-Wilk test for
normality), nonparametric comparison of groups (the Mann-Whitney U test),
principal component analysis (PCA), hierarchical clustering (Ward's D2 method),
the K-means method with validation, and permutational multivariate analysis of
variance (PERMANOVA).
According to the data analysis, the rivers differ reliably in 65.6% of the
parameters. A triad of specific markers of oil production activity was identified:
chlorides (6.2-fold excess), molybdenum and strontium (both markers of formation
waters), as well as cadmium and lead. Principal component analysis showed a clear
separation of the point clouds of hydrochemical data of the two rivers (overlapping
less than 10%). The main contribution to PC1 is made by barium (9.63%), and to
PC2 – by BOD5 (10.73%). Clustering identified three levels of pollution:
conventionally clean water (n≈71), moderate pollution (n≈140), and high pollution
(n≈16, all samples from the Sheshmа River). PERMANOVA confirmed the
statistical significance of the differences (F₁,₂₂₀=19.039, p=0.001).
Based on a multivariate analysis, five specific indicators of oil production
activities were identified and are recommended for implementation in industrial
and environmental monitoring systems. This work has important practical
applications for identifying pollution sources and establishing liability for
environmental damage in oil-producing regions, as well as theoretical significance
for the development of multivariate methods in environmental diagnostics.

• PDF (Russian)