Pacific cirrus above stratocumulus
7 December 2020 12:50-23:00 UTC
The new NIR1.3 channel on GOES ABI helps to detect thin cirrus clouds above low-level clouds.
07 December 2022
03 February 2021
By Jochen Kerkmann (EUMETSAT) and Daniel Lindsey (NOAA)
In December 2020, CIRA posted a very nice case of 'Pacific cirrus above stratocumulus' on their GOES loop of the day website. The GOES-16 nine-hour animation is shown in Figure 1. Note: how high-level cirrus and low-level stratocumulus move in different directions.
The product displayed is the so-called Day Snow/Cloud Layers RGB product, which combines information from six different bands on the GOES ABI to help distinguish clouds from snow and ice. Unlike other RGB snow/ice products, snow is portrayed in a more intuitive colour – white (for example see GOES-16 animation of snow in the Arizona desert on 27 January 2021). Also, this product discriminates very well between low clouds (in yellow) and high clouds (in pink/magenta). The algorithm uses combinations of reflective and infrared bands, that includes a cirrus mask from the 1.38μm channel (band 4) , and a normalized difference snow index (NDSI). See the quick guide of the RGB product.
Like all near-infrared bands, the ABI 1.38µm band detects reflected solar radiation (from clouds or land surfaces, see NIR1.3 quick guide. However, at this spectral wavelength, solar radiation is strongly absorbed by water vapor. Thus, in a humid atmosphere, only high clouds are seen. In a very dry atmosphere, low clouds (or snow on mountains) can be seen.
Figure 2 shows the NIR1.3 band at 18:00 UTC over the Pacific and western South America. Two different enhancements are shown: a range of 0 to 60% to give an overview of the high cloud situation (mainly thick high clouds are seen), and a range of 0 to 10% (with a Gamma of 2.5), which over enhances thick high clouds, but strongly improves thin cirrus detection. In particular, the thin cirrus over the Pacific becomes better visible in this enhancement. However, the cost of this enhancement is that other features (in this case lower level clouds over the Pacific, and snow in the Andes) get enhanced too and could be confused with thin cirrus.
GOES-17 NIR1.38 band comparison
Figure 2: GOES-17 NIR1.38 band, 7 December 18:00 UTC. Left 0-60% enhancement, right 0-10% enhancement
Figure 3 gives a comparison of the corresponding Cloud Type RGB (based on solar channels) and the Dust RGB (based on IR channels). The Cloud Type RGB shown here uses the NIR1.3 band on red (range 0-10%, Gamma 2.5), VIS0.8 band on green (range 0-100%) and NIR1.6 band on blue (range 0-60%). In the Cloud Type RGB thin cirrus clouds appear with red colour, thick high clouds are yellowish and low level clouds are cyan/green (see quick guide of this Cloud Type RGB product). Both RGB products are about equally good in detecting thin cirrus clouds, but the colour contrast to low level clouds is better in the Cloud Type RGB, we think.
GOES-17 image comparison
Figure 3: Comparison of GOES-17 Cloud Type RGB (left) and Dust RGB (right), 7 December 18:00 UTC.
Figure 4 shows a sequence of hourly Cloud Type RGB images, from 13:00 to 23:00 UTC. The area shown here is a bit larger than in Figure 1, it includes western South America (Chile and Peru). In addition to the cirrus and the stratocumulus clouds over the Pacific, the diurnal convection over the Peruvian Andes can be observed. Note that, in the evening hours, all high clouds become reddish, i.e. the distinction between thick and thin high clouds is no longer possible (one of the known limitations of this product for low Sun illumination).
