Beyond AOD – Remote Sensing of Desert Dust Physico-Chemical Properties from Infrared Satellite Observations

Desert dust is characterized by strong silicate absorption bands located within the atmospheric window region in the terrestrial infrared (TIR) between 8 µm and 12 µm. These absorption bands and the corresponding optical properties (extinction efficiency, single scattering albedo, scattering phase function) have very specific spectral shapes for different silicate minerals, modulated by the particle size and shape. The asphericity of desert dust particles strongly affects the absorption band characteristics, for example due to surface wave modes for small particles. The use of the correct particle shape model significantly increases the spectral correlation between simulated dust optical properties for typical minerals and corresponding laboratory measurements. This information is exploited for infrared satellite remote sensing as well. The Infrared Mineral Aerosol Retrieval Scheme (IMARS) has been developed for the Infrared Atmospheric Sounding Interferometer (IASI) on board the European Metop satellite series. It exploits the characteristics of the spectral shape of the radiance measurements to infer dust (and ice cloud) properties in a probabilistic approach. A set of four combined brightness temperature differences from three pseudo-channel bands are generated from hyperspectral IASI observations. Those bands are located at similar positions as in narrow-band imaging instruments. The strong advantage of the hyperspectral signal from instruments like IASI is the capability to minimize the influence of disturbing gas absorption lines within these bands.

Comparing forward simulations with different dust particle size distributions and mineralogical mixtures with the observations allows for selecting the most appropriate dust model. Instead of selecting one specific model from a look-up table, the IMARS approach uses the concepts of mutual information and signal entropy for creating a probability density distribution (PDF) over the different dust models. Such an approach has the advantage that the number of independent signals (variables) for each observation is directly provided by the retrieval itself. For desert dust this number typically ranges from 2.5 to 4.0 depending on the characteristics of the observed dust plume. Consequently a lot more information beyond Aerosol Optical Depth (AOD) can be retrieved from these measurements.

The dust model PDF directly provides information about dust particle size distribution (PSD) and mineralogical composition as well as a particle shape factor. The dust PSD is characterized by effective radius and mass-weighted mean diameter. The former is sensitive to the cross sectional area, whereas the latter represents the particle volume. PSD and mineralogical composition (also defining the particle shape model) are also used to dynamically calculate a conversion ratio for AOD from 10 µm  to 0.55µm.

Moreover, the effective dust emission temperature is retrieved as well. This radiometric quantity can be converted into dust layer altitude by interpolation of atmospheric temperature profiles.

Within the Aerosol_cci project of the European Space Agency (ESA) IMARS dust observations with IASI starting in 2007 (start of the first Metop satellite) until present are produced and made publicly available. These include all above mentioned dust variables. This dataset is used to calculate a climatology of physico-chemical dust properties and to track significant dust events for identification of the corresponding dust sources.

The IMARS algorithm can generally also be applied to narrow-band imagers. The information content is slightly reduced but still sufficient to retrieve dust information beyond AOD. Examples from SEVIRI (Spinning Enhanced Visible and InfraRed Imager) and VIIRS (Visible and Infrared Imaging Radiometer Suite) will be shown to demonstrate the potential for global dust monitoring with upcoming geostationary satellites such as HIMAWARI, GOES-R and the third generation of Meteosat (MTG); these will also significantly increase the spatial resolution of quantitative desert dust observations from space.