During several days around mid-November 2021 a layer of haze was covering large part of India stretching into the Arabian Sea.
15 December 2022
13 December 2021
By Ivan Smiljanic (CGI), Natasa Strelec Mahovic (EUMETSAT) and HansPeter Roesli (Switzerland)
In the autumn months, October and November, a haze 'cloud' can often be seen over India and the adjacent sea. There can be multiple sources of the haze: agricultural fires, dust particles, urban pollution, traffic pollution.
Figure 1 shows the haze in the Suomi NPP True Color RGB image. The haze is clearly seen over both the land and the sea. The same is also visible in Figure 2, in the RGB combination of channels 2.25, 0.8 and 0.6 of Suomi NPP VIIRS. The haze appears in grey to brown shades and is, therefore, better visible over the sea than the land due to colour contrast with the dark sea surface.
At the same time, looking at SEVIRI imagery from the Meteosat-8 (IODC) satellite in Figure 3, the haze can barely be detected, perhaps only over the sea.
Comparing the imagery from SNPP and Meteosat-8 (Figures 1 and 2 versus 3) highlights the benefit of higher spatial and spectral resolution in detecting aerosols and particles of different sizes. The SEVIRI instrument has a 3km resolution in nadir, whereas the NPP VIIRS instrument images have 1km resolution. The VIS0.45 and VIS0.55 micron channels, available from VIIRS, are proving to be very valuable in particle size detection, especially for a small-size aerosols. Both of these properties of VIIRS images mimic the capabilities that will be available with future Flexible Combined Imager (FCI) sensor on Meteosat Third Generation satellites (MTG).
The ability to detect haze (and other aerosols in the near-IR and visible spectral region, very much depends on the angle between the source of radiation (Sun), observed target (aerosols) and observing instrument (satellite) (Figure 4). The bigger the angle the more 'forward-scattering' from the aerosols. This also applies to back-scattering, though it is less intense than the forward one. The fact is that scattering of light on the very small aerosol particles (Rayleigh scattering,) and on the particles with comparable or bigger particle size than used wavelength (Mie scattering), has properties such that the forward (and backward) scattering is much stronger than the side scattering. Illustration of scattering angle dependence for Rayleigh and Mie scattering is given in Figure 5.
This is apparent when comparing the same products from two of the same instruments/satellites with a different orbit position. When comparing the Natural Colour RGB form Meteosat-11 (0°) and Meteosat-8 (41.5°E) it is clear that the former satellite is much better at capturing aerosol scattering. (Figure 6)
In Figure 6 we changed the satellite angle, with a fixed Sun angle. The following time animation (Figure 7) demonstrates the power of forward-scattering with a changing Sun angle, and fixed satellite angle (Meteosat-8). Though it appears that local aerosol concentrations are dropping after the sunrise, as time passes, higher Sun angles actually contribute to reduced 'detection power' by the SEVIRI instrument.
Often, especially if one is interested in observing features other than aerosols in the satellite imagery, so-called Rayleigh correction is performed. Such correction removes the signature of Rayleigh scattering, and it has to take into account the solar zenith angle and satellite viewing geometry. Such a correction can also lead to considerable reductions in error for many satellite products, e.g. cloud optical thickness. Examples of such correction, i.e. comparison between original and corrected satellite imagery, for both Metosat-8 and Himawari-8 satellite, are given below.
Note how the correction is progressively more apparent from Figure 8 to Figure 10. This is because Himawari-8 has a higher satellite viewing angle over the observed region (v Meteosat-8), but also due to channels of shorter wavelengths used to construct the True Colour RGB (v Natural Colour RGB)