Dust aerosols have an impact in the thermal infrared wavelengths that we can use to detect aerosols over desert
surfaces. To retrieve the aerosol properties over land, we have to account for the surface contribution. The surface
radiation depends on the skin temperature, which is characterized by a strong diurnal variation. Therefore, it is
better to use the surface emissivity, which we assume constant over a time span of 24 hours.1
The surface emissivity is based on clear sky observations that are corrected for atmospheric extinction and
emission. The clear sky image is a composite of pixels that is characterized by the highest brightness temperature
of the SEVIRI channel at 10.8μm. Due to the lower temperatures of clouds and aerosols we can assume that the
selected pixel values are obtained for a clear sky day.
We use a forward model to simulate the thermal infrared radiation transfer in the dust layer. The apparent
surface radiation in the presence of aerosols is calculated as a function of the geometric angles, the surface
emissivity, and the aerosol optical depth (AOD). This is stored in lookup tables (LUT) that are inverted to
retrieve the AOD from the observed apparent surface radiation.
The retrieval algorithm consists of firstly, processing the clear sky image and computing the surface emissivity,
secondly, processing of the instantaneous image and computing the apparent surface radiation, and thirdly,
selecting the corresponding LUT and retrieving the AOD that matches to the observed apparent surface radiation.