Feature extraction of climate variability, seasonality, and long‐term change signals in persistent organic pollutants over the Arctic and the Great Lakes

Abstract

We explored the interactions between climate variability and measured persistent organic pollutants (POPs) data in the frequency domain. The spectrum analysis identifies a power period of 1.5 to 7 years in sampled POPs data in the Great Lakes (GL) and Arctic in the 1990s to 2000s, coinciding with the power period of two climate variabilities, the North Atlantic Oscillation and the El Niño–Southern Oscillation. Results also reveal a decadal power period in monitored POPs time series, which is associated with air temperature and climate indices, demonstrating the connection between the dominant long‐term trend in sampled POPs data and climate change. We used a “cocktail party effect” and Fast Independent Component Analysis approaches to blindly separate and isolate the signals in POPs time series from different dynamic and environmental processes in the Arctic and the GL. Among seasonal, interannual, and decadal or longer‐term signals extracted from sampled POPs concentration data, the climate signals in hexachlorobenzene in the GL are associated with air temperature, El Niño–Southern Oscillation indices, and North Atlantic Oscillation (NAO). The signals in α‐hexachlorocyclohexane (HCH) and polychlorinated biphenyls observations at three Arctic sites correspond to a Niño 3.4 and the NAO. The other two signals are associated with the longer term and seasonal cycle of POPs. We show that the seasonal cycle and climate change are important factors, buffering the declining trend of α‐HCH. The signals feature extraction from POPs historical data provides a tool to deconstruct the POPs time series and enhance our understanding of the potential effects of climate variations on POPs environmental cycling.

The Tensor‐based Feature Analysis of Spatiotemporal Field Data With Heterogeneity

“…The original data are the multiyear average data, they are susceptibly influenced by many common climate‐forcing factors (such as terrain or climate events in different periods of the year), showing the significant heterogeneity in temporal behavior (Zhao et al, ). In the absence of prior knowledge, the original data are partitioned into 12 subtensors according to 12 months, and then they are reorganized based on the proposed data reorganization strategy.…”

Nonlinear PCA for Spatio-Temporal Analysis of Earth Observation Data

“…The challenge is to derive spatial and temporal components that summarize the information content of the data cubes, while being physically meaningful and interpretable. Traditional techniques of feature extraction, such as the removal of mean seasonality, temporal trends, parametric fitting or harmonic decomposition are practical and commonly used [3]- [7]. However, they require prior knowledge and assumptions, and therefore impose expected relations that could not necessarily be found in data.…”