Channel Differencing and RGB Issues Over Desert Locations…

Recently, I had the opportunity to travel to the Tucson NWS office and work with forecasters there concerning a number of experimental data sets transitioned by the SPoRT group.  Primarily, this involved the SPoRT LIS, GPM Constellation and IMERG, and NESDIS QPE data sets.  However, I also had the opportunity to see how other products were being utilized by forecasters.  While taking a look at the Nighttime Microphysics RGB image, I was initially perplexed by the apparent presence of fog and low clouds in parts of the desert southwest.  The first image below is a 4-panel image from AWIPS, showing the Longwave (LW) and Shortwave (SW) IR, the LW-SW IR channel difference, and the Nighttime Microphysics RGB from the VIIRS instrument on the morning of Sept 23rd.

Image 1. Suomi-NPP VIIRS imagery valid 0915 UTC 23 Sep 2015, Longwave IR (upper left), Shortwave IR (upper right), LW-SW IR channel difference (

Image 1. Suomi-NPP VIIRS imagery valid 0915 UTC 23 Sep 2015, Longwave IR (upper left), Shortwave IR (upper right), LW-SW IR channel difference (“fog product”, lower left), and the Nighttime Microphysics RGB (lower right).

The difference in brightness temperatures between the LW and SW IR channels in parts of SW Arizona, SE California and areas of NW Mexico around the Gulf of California, results in relatively large positive values.  Notice the yellow colors that appear in these areas in the channel difference imagery (image 1, lower right), and the corresponding appearance of white-aqua colors in the Nighttime Microphysics RGB (the 10.8-3.9 channel difference represents the green color component of the RGB recipe).  For a forecaster accustomed to looking at these imagery in other parts of the country (and those will less sandy surfaces), these channel difference values and colors in the RGB would suggest the presence of low stratus and/or fog.  However, no clouds or fog were present in those locations during the morning.  You can, however, see some low clouds in portions of central and eastern New Mexico, as indicated by the brighter white-aqua colors.

So, what is going on here?  Well, as eluded to above, it’s the presence of dry sand.  The image below (courtesy of COMET) shows the IR emissivity over several different surface features: tree leaves, red clay, dry sand, and water.

Image 2. IR emissivity vs. wavelength of several surface features, including tree leaves, red clay, dry sand, and water.

Image 2. IR emissivity vs. wavelength of several surface features, including tree leaves, red clay, dry sand, and water.  (image courtesy of COMET)

Notice that the emissivity over dry sand changes fairly substantially through portions of the SW and LW portion of the spectrum, and is lower at 3.9 µm than at 10.8 µm.  The channel difference between 10.8 and 3.9 µm will result in positive values (given clear sky conditions of course) over dry sandy areas, thus mimicking the presence of low clouds and/or fog, as would be the interpretation in other areas.  The next image below demonstrates the LW and SW IR brightness temperatures and differences, along with the Nighttime Microphysics RGB, as sampled over a clear, dry sandy area.

Image 3. Suomi-NPP VIIRS image from 0902 UTC 25 Sep 2015

Image 3. Suomi-NPP VIIRS image from 0902 UTC 25 Sep 2015, LW IR (upper left), SW IR (upper right), LW-SW IR channel difference (lower left), and the Nighttime Microphysics RGB (lower right).

Notice the substantial resulting green color contribution in the Nighttime Microphysics RGB (lower right in above image).  These colors are very similar to colors that would be indicative of fog and other low cloud features as they traditionally appear under similar temperature conditions in other areas outside of dry, sandy areas (image 4 below).

Image 4. Nighttime Microphysics image depicting fog and low clouds (white-aqua colors) in portions of the southern and central Appalachian region.

Image 4. Nighttime Microphysics image depicting fog and low clouds (white-aqua colors) in portions of the southern and central Appalachian region.

One thought on “Channel Differencing and RGB Issues Over Desert Locations…

  1. Fascinating, Kris. Thanks for sharing this important application, reminding all that emissivity is a function of land surface characteristics. PN

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s