Use and impact of PCA-reconstructed radiances for atmospheric composition applications

The IASI mission has been developed with the objective to provide high-quality radiance spectra and global measurements of atmospheric parameters, for accurate Numerical Weather Predictions (NWP) and improved climate and atmospheric composition (AC) monitoring. This led to the design of an advanced hyperspectral infrared instrument, featuring, at Level1C a large spectral range of the infrared without gaps (645-2760 cm-1), a fairly high spectral resolution (0.5 cm-1) and a low noise. These Level1C radiances have been exploited extensively in the last decade for applications in air quality, climate, tropospheric and stratospheric chemistry. There is in fact an impressive series of atmospheric trace gases that are retrieved from IASI Level1C, including several that display very small spectral signatures or others that appear only around localized pollution sources or in volcanic and fire plumes. Because of the potential loss of information coming with principal component analysis (PCA) –especially for rare signals- the scientific community has, until now, favored the use of the native Level1C over PCA-reconstructed radiances, despite the reduction of noise that the latter allow. In this presentation we will show results obtained during a dedicated study with IASI spectra over a decade, with the objective 1. to evaluate how well the IASI Level1C can be reconstructed from the eigenvectors routinely produced at EUMETSAT, and to document the added value of the so-called dynamic hybrid over the static core base eigenvectors for rare or extreme situations. 2. To evaluate to what extent the use of reconstructed radiances impacts the retrieval of trace gases, in background or extreme situations.The results concern ozone (validation of the retrieved profiles against ozone sondes), carbon monoxide, ammonia and sulfur dioxide in volcanic plumes. Based on these results, recommendations for using reconstructed radiances in an operational processing framework are made. This is important considering that reconstructed radiances will become the basis of the operational Level1 dissemination for future hyperspectral IR sounding missions, starting with IRS-MTG.