NO2 vertical column density estimation from interferograms captured by a snapshot interferometric imaging spectrometer

Nitrogen dioxide is an atmospheric gas of major impact on climate change and air quality and its monitoring through its detection and quantification is essential to control its impact on the environment and health.

The detection and estimation of trace gases are limited by cost of the acquisitions devices and spectro-temporal resolution of the acquisitions. Conventional imaging systems are the result of a trade-off in terms of size, and spectral and spatial resolutions.

In order to overcome these technological limitations, a new device known under the patent of Imaging Spectrometer on Chip (ImSPOC) allows for real-time acquisition and a significant spatial resolution. As its volume about the size of a matchbox, it has the potential to become a base brick for nano-satellites, drones, or ground-based measurement platforms. The ImSpoC device is based on an array of Fabry-Perot interferometers of diffrent thickness mounted over a high-sensitivity CCD imaging detector (a matrix of photodiodes). As ImSPOC performs a division of the field of view (with a matrix of
micro lenses coupled with the interferometers), a typical acquisition consists in a matrix of sub-images which can be recombined in order to form a single hyperspectral image of the observed scene in which each pixel yields an interferogram.

When ImSPCO is used as a spectrometer, a common processing involves reconstructing spectra from interferograms as an inverse problem. This operation is importatnt since the commonly used techniques, such as Differential Optical Absorption Spectroscopy (DOAS) rely on light spectra instead of interferograms.

This work explores a way to adapt these techniques directly on the interferograms captured by ImSPOC is explored by relying on an optical model of the instrument. As spectra and interferograms are linked by a Fourier-like transform, the interferograms exhibit the periodicity of the incident light spectrum. In this work we propose to use an optical filter to isolate a wavelength range where the absorption cross-section of NO₂ is strongly periodic and not correlated with that of other trace gases. We expect that the correlation between a difference of interferograms and the Fourier transform of the (filtered) absorption cross-section of the target gas is proportional to the difference of slant column densities of the targeted gas, here NO₂. 

Some experimental results were obtained by processing ImSPOC acquisitions over several hours at sunrise, noon, and sunset in two configurations: zenith light (the sensor being oriented towards the zenith), and direct light (the sensor being directed towards a surface of material with high diffuse reflectance). In order to validate the results obtained by processing the ImSPOC acquisitions, data from a conventional diffraction grating based spectrometer were used, providing reference measures for the air-mass factors and allowing for a comparison with a regular DOAS method.