Separation of DOAS measurements data by independent component analysis

Differential optical absorption spectroscopy (DOAS) performs optical measurements of gaseous air pollutants concentrations by open path monitoring, which is a "non-invasive" technique. Concentrations of trace gas species in the light beam's path (1 km) are determined simultaneously by the analysis of the measured differential absorption transmission spectra. We proposed the use of independent component analysis for the separation of trace gas spectrum from transmission spectra. Laboratory measurements (like NO2, SO2, O-3 and HONO) were used as differential absorption cross sections for trace gases. Because the absorption wavelength range of trace gases is different, we can use the sparsity of reference absorption cross sections. If reference spectra have super-gaussian distribution, nongaussian sources can be recovered from their mixtures using spatial structure by spatial independent component analysis (sICA). And we imposed the prior information of reference spectra on ICA. Simulation results show that DOAS based on ICA is well suited for a simultaneous separation of atmospheric trace gases.