Analysis of Thermal Emission Spectrometer data using spectral EOF and tri-spectral methods

We introduce two new techniques in analyzing martian spectrally resolved radiance data obtained by the Thermal Emission Spectrometer (TES): spectral empirical orthogonal function (EOF) analysis and the tri-spectral algorithm. Spectral EOF analysis allows us to obtain the variability of spectra and associated temporal and spatial patterns. The case study with TES 20° S–20° N data shows that the first principal component (PC1) dominates the total variance and is associated with surface or near-surface brightness temperature variations. The PC2 is associated with atmospheric variability, and a negative correlation between dust and ice absorptions can be clearly seen over many regions. The annual cycle is a major component of the PC1 temporal patterns. The fingerprint of the dust storm can be clearly seen in the PC2 temporal patterns in most areas except the highlands. Spectral EOF can be used for validation of the variability of martian GCMs. The tri-spectral algorithm is based on the differences between three bands (dust, ice and a weak CO2 absorption band) to distinguish spectra sampled in different situations: water ice cloud, dust, and surface anisothermality. We use a line-by-line radiative transfer model coupled with multiple scattering to investigate the sensitivity of this algorithm to dust and ice optical depth as well as surface emissivity. The comparisons between results of this algorithm and the TES teamʹs retrieved dust and ice opacity are consistent over all studied periods except during the peak of the dust storm. Our algorithm is complementary to the more sophisticated TES retrieval and can be used to screen large amounts of data to get an overview.