Fast Quantification of Air Pollutants by Mid-Infrared Hyperspectral Imaging and Principal Component Analysis

Area of Science:

• Atmospheric Spectroscopy
• Environmental Monitoring
• Data Analysis

Background:

• Accurate quantification of atmospheric pollutants is crucial for environmental monitoring.
• Fourier-transform spectroscopy provides detailed spectral information but can be computationally intensive for retrieval.

Purpose of the Study:

• To develop and validate a faster and more accurate method for retrieving atmospheric pollutant concentrations using spectral data.
• To assess the efficacy of Principal Component Analysis (PCA) in enhancing spectral retrieval from mid-infrared transmittance spectra.

Main Methods:

• Acquisition of mid-infrared transmittance spectra (1850-6667 cm-1) for methane, nitrous oxide, and propane using an imaging Fourier-transform spectrometer.
• Retrieval of column densities (Q) and temperature (T) by fitting experimental spectra with theoretical spectra from the HITRAN database, incorporating a radiometric model.
• Application of Principal Component Analysis (PCA) to experimental and simulated spectral data to reduce dimensionality and improve signal-to-noise ratio.

Main Results:

• Two principal components were sufficient to reconstruct gas spectra with high fidelity.
• PCA processing improved signal-to-noise ratio and spatial resolution, leading to more uniform retrieval.
• A computational time reduction exceeding a factor of one thousand was achieved with improved accuracy using PCA for retrieval.

Conclusions:

• Principal Component Analysis (PCA) is a highly effective technique for accelerating and enhancing the accuracy of atmospheric pollutant retrieval from spectral data.
• PCA offers a significant advantage in processing spectral data, enabling faster and more precise quantification of atmospheric gases.
• The study suggests that retrieval can be further simplified by expressing temperature and column density as quadratic functions of the first two principal components.