The field dataset including physico-chemical characteristics of Khibiny alkaline massif groundwater aquifer was analysed for the first time by applying chemometric methods to explore the relationships between investigated parameters in both time and frequency domains. Elevated Al concentration in water of this aquifer is described as a serious health concern when used for the household water supply. The time series of 12 physico-chemical parameters were examined by using Pearson correlation, multiple linear regression and spectral analysis based on fast Fourier transform (FFT) algorithm. Computed pairwise correlation coefficients matrix revealed that pH values, NO3-, SO42-, Cl- and TDS correlate with Al concentrations at a statistically significant level (α-level = 0.05). The multiple regression model including Cl-, NO3- and pH explains up to 54% of Al concentrations temporal variation in groundwater. Fourier coefficients, power spectral density (PSD) and cumulative spectral power (CSP) were calculated for Al, pH, NO3- and Cl- time series to identify and analyse the strength of their variations as a function of frequency. Calculations revealed the spectrums of these variables include three main frequency bands corresponding approximately to 5-7, 13-17 and 20-34 month periods. The fluctuations within these bands contribute mostly to the total temporal variation of Al, Cl-, NO3- concentrations and pH values.
Keywords: Aluminium; Fourier transform; chemometrics; groundwater; multiple regression; spectral analysis.