Sunglint over the Arabian Sea observed by SEVIRI IR channels.
16 June 2022
24 April 2011
By HansPeter Roesli (EUMETSAT)
Searching in Meteosat-9 Dust RGBs of for a possible rain event along the Omani coast, has led to the rather lucky discovery that the Dust RGB composite may strongly enhance the IR part of sunglint. Sunglint, while also present in the IR part of solar radiances and noted in literature for disturbing SST retrievals (for a couple of references see below), usually goes unobserved in standard IR imagery.
With hindsight, under excellent sunglint conditions (close to pure reflection over calm sea) the Dust RGB can be expected to show off distinctly as deep blue blotches/freckles (warm sea temperature augmented by the reflected IR radiation from the Sun), in the absence significant contributions from the red and green colour beams (clear sky and very similar emission from water in all three IR channels involved).
The animated sequence of Dust RGBs on 24 April between 01:30 UTC and 03:45 UTC shows the blue freckles travelling prominently off the Omani coast across the Arabian Sea. Of course, the sunglint also shows up in the individual IR channels, but only when adequately enhanced, e.g. in IR10.8 in the temperature range 290–300K, whereas the solar channels are being massively blinded (see 3:00 UTC images). The sunglint appeared not only on 24 April, but for several days around that date, as illustrated by the Dust RGB imagery between 01:00 UTC and 03:45 UTC on 22 to 26 April.
Looking closer into the radiances of SEVIRI IR channels 5 to 11, the window channels IR8.7/10.8/12.0 logically show the strongest signals. In the channels affected by absorption WV6.2 shows a weak signal, whereas in channels WV7.3 and IR9.7/13.4 no sunglint is apparent.
Comparing the pixel marked 'pixel UR' in IR12.0 (see strong zoom of IR window channels), which is the strongest pixel at 03:00 UTC (and is already off the sunglint at 03:15 UTC), the strongest on/off sunglint signal is coming from the IR8.7 channel (34.8mW/m2/cm-1, i.e. about 25K), followed by the IR10.8 channel (13.3mW/m2/cm-1, i.e. about 8K) and the IR12.0 channel ( 5.3mW/m2/cm-1, i.e. about 3K). The decrease is compatible with the diminishing solar contribution with increasing wavelength.
From the second references below it appears that the sunglint disturbance of SST retrievals from the AATSR instrument on ENVISAT only is in the order of some tenths of degrees. If this is true the sunglints reported here appear to be exceptional and might warrant some further investigation.
Cox, C., & W. Munk. 1954: Statistics of the sea surface derived from Sun glitter. J. Mar. Res., 13, 198–227.
Zavody A. M., & A. R. Birks., 2003: Sunglint contamination in the ATSR-2 data: comparison of observations and values calculated from the Measured 1.6-µm reflectivities. J. of Atmos. And Oceanic Tech., 21, 787–798.