Application of Hilbert-Huang Transform Method for Analyzing Toxic Concentrations in the Niagara River

Abstract

A novel application of the Hilbert-Huang transform (HHT) method for the analysis of nonstationary and nonlinear time series is proposed in this paper. The HHT is introduced to the field of water quality analysis both to evaluate trends and to assess the cause-effects relationship between the system under study and external physical factors. The HHT combines two distinct analytical methods: the empirical mode decomposition (EMD) to decompose observed series into independent intrinsic mode functions and the Hilbert transform to transform these time-dependent functions into time-frequency functions. A practical application of the HHT method for data analysis is presented herein for four time series in the Niagara River: flow, water temperature, and incoming concentrations of two polycyclic aromatic hydrocarbons. It is shown that the EMD improves the spectral representation of the original time series, enabling a better detection of periodic trends and dominant time scales. It is concluded that the ability of the HHT of decomposing the signal based on the data and transforming it into the frequency-time domain provides insights into time series characteristics that could improve the modeling effort on the Niagara River.