During the last years there has been considerable interest in the development of efficient methods for the detection of subsurface contamination and monitoring pollution in the field. Traditional methods of characterizing the contaminated groundwater and soils involve sampling and laboratory chemical analyses, which are costly and time-consuming processes. In addition, the risk of samples contamination during sampling, transportation and analysis is always substantial. So, an inexpensive method must be used, enabling preliminary inspection for chemical changes in the measured samples, related with the pollutants. Dielectric spectroscopy has been proposed by many researchers as a promising tool and experiments have been carried out to establish the sufficiency of this technique to identify subsurface contamination and its sensitivity to different kinds and concentrations of organic or inorganic pollutants. In the present contribution, dielectric measurements of leachates-contaminated soil samples were carried out in the frequency range of 10mHz to 1MHz by means of a high resolution broadband spectrometer (Novocontrol Alpha-N Analyzer) and time series of dielectric permittivity produced at discrete measuring frequencies. The analysis of the dielectric spectra is not always easy and various empirical and theoretical models have to be taken into consideration for the interpretation of the data. So, an innovative attempt took place to study contamination by the use of dielectric permittivity time series on various frequencies applied on pure and contaminated soil samples. For each produced time series, power spectrum density (PSD) is calculated by means of Wavelet Transform (WT) since stationarity of the time series is not verified, as required for use Fourier transform. PSD is calculated in both ways: using wavelet coefficients and by denoising periodogram. PSD time series results, which were analyzed by means of wavelet transform calculations, reveals an important diversification according to the slope spectrum between contaminated and uncontaminated samples. This promising result can guide us in a recognition scheme where we may identify polluted or contaminated materials using dielectric measurements and examining the slope of the wavelet based calculated PSDs from dielectric permittivity time series.