GOES-18, the third satellite in the GOES-R series, launched on March 1, 2022. First light ABI imagery from 89.5W was released on May 11 and can also be found here and here. GOES-18 completed its drift to the ~GOES-West position on June 6. Early GOES-18 ABI imagery from this position can be found here and here and in the article/video here. The cooling system on the GOES-18 ABI was redesigned in order to prevent the issues that plague the GOES-17 ABI during certain time periods (GOES-17 LHP info). The redesign was successful, as the cooling system on GOES-18 is working as expected. GOES-18/GOES-17 interleave begins on August 1, meaning NWS users will be viewing GOES-18 imagery in AWIPS (instead of GOES-17) through September 6 (during the ABI Warm Period).
There is a minor artifact apparent in the GOES-18 Channel 07 3.9 um “shortwave IR” channel and associated multispectral products (band differences and RGBs). This artifact is not related to the cooling system, which is working correctly on GOES-18 ABI. Since this artifact is noticeable in some imagery products at times, it is important to acknowledge its appearance in the single-channel imagery as well as in multispectral imagery products. Given its appearance in imagery as a dynamic series of vertical striping, as you will see below, the artifact has been dubbed the “Barcode Artifact”. A Satellite Book Club Webinar discusses the issue in detail (link). This post will introduce the impact of the “Barcode Artifact” on the various imagery products available in NWS AWIPS that leverage the 3.9 um channel. The GOES-18 Transition to Operations page here also includes information on the Barcode Artifact (Downloadable PowerPoint). All of the imagery in this blog post was created in AWIPS from the CMIP files.
Summary up front: The “Barcode Artifact” is most apparent in band differences involving the 3.9 um band and in RGBs using those band differences. Most notable to NWS users is the presence of the artifact in the Night Fog Difference and Nighttime Microphsyics RGB. The artifact is most noticeable in cold scenes such as overnight, and in cold cloud tops. The vertical striping is very subtle in the 3.9 um single band imagery and associated RGBs (FireT RGB). The artifact appears to mostly serve as an annoyance, and should not impact a forecasters ability to analyze the imagery and diagnose relevant features.
The “Barcode Artifact” is not easily seen in single-band 3.9 um imagery. Therefore, routine day-to-day monitoring of the single-band 3.9 um imagery, such as for wildfire hot spots, when using typical ranges and colormaps, will not be impacted. The example in Figure 1 is of the single-band 3.9 um imagery over the western US from 12Z on July 23 to 12Z on July 24 with the AWIPS default rainbow colormap and default range of -109C to 127C. Compare the GOES-18 example in Fig 1 with the GOES-17 example in Fig 2. The imagery appears almost identical, but a keen eye can spot differences in the cold cloud tops over the southeast portion of the scene, as well as over the ocean during the overnight periods, with shifting and subtle vertical striping present in the GOES-18 imagery and not GOES-17.
Some may prefer a grayscale colormap when viewing 3.9 um imagery (black = warm to white = cold BTs), such as that shown in Figure 3, still with the default range. In this example, the only apparent artifact is again barely observed in the cold storm tops to the southeast and over the ocean overnight.
Using the same grayscale colormap, but spreading the range of colors across a smaller range of values (0C to 32C), the artifact finally becomes more apparent (Fig 4). The series of inconsistent vertical striping is detectable to the human eye at night across the full scene (start and end of animation), but especially offshore and in clouds. In comparison, the artifact is not present in similarly stretched GOES-17 imagery (Fig 5).
The Fire Temperature RGB leverages the single-band 3.9 um channel for hot spot detection. A day/night example over wildfires in Idaho reveals no noise that would hinder a forecasters ability to diagnose hot spots or other features (Fig 6). In this RGB, only values greater than 0C in the 3.9 um band contribute to the red component (relatively warm BTs), hence a lack of striping.
Another vital role of the 3.9 um channel is for low cloud and fog detection, when combined with the longwave IR clean window (10.3 um) channel in the “Night Fog” difference. This is where the the artifact becomes more noticeable in areas. In this blog post, the 10.3 um – 3.9 um difference is used with the AWIPS default fogdiff_blue colormap. Positive values, or light blue to dark blue, represent liquid cloud tops (lower-level clouds and fog), shades of gray to black are ice cloud tops (higher-level clouds).
During the same period and in the same location as in Fig 1, an array of vertically oriented stripes or bands is readily apparent shifting across the scene in the Night Fog Difference (Fig 7). The artifact is most apparent at night (start and end of loop), and in both cloud tops and clear skies over land and water. Despite the artifact, the various cloud features across the domain are still easily diagnosed and differentiated by the professional analyst. If users are aware of this artifact, other than being a slight annoyance, there should be little to no impact on ones ability to perform cloud analysis. In some instances, a cloud feature falling on the gradient of colors in a given colormap may appear to flash in and out artificially, particularly in the coldest of scenes (Alaska example below). Animations of images help the user to comprehend close to the actual cloud location/appearance.
Figure 8 includes a corresponding animation of the Night Fog difference from GOES-17. Notice the presence of horizontal striping during the nighttime hours, which are associated with the cooling system issue. The vertical “Barcode Artifact” apparent in the previous GOES-18 example is not detectable in the similarly scaled GOES-17 imagery.
Of course, the fog difference is an important component to a couple of AWIPS RGBs, namely the Nighttime Microphysics RGB. The vertical striping appears similarly in the RGB as it does in the fog difference, as is shown in Figure 9 zoomed in over southern CA and adjacent waters during the overnight period. GOES-17 imagery of the same product and timeframe does not show the artifact (Fig 10).
The more advanced “Geocolor” product also leverages the Night Fog difference (for low clouds) at night to highlight low clouds and fog as blue. The vertical striping appears similar across the region of low clouds as in previous examples, with analysis of low clouds still able to be accomplished (Fig 11).
The Day-Snow Fog RGB also incorporates the (reverse) Fog Difference (Fig 11-2). Vertical striping is not observable in this example, which isn’t surprising given it is a daytime and low-level application.
This fog difference is also included in the Day Cloud Convection RGB. In this case, the difference is leveraged during the day to differentiate small ice particles in the cloud top as a sign of active updraft regions, yellow in the RGB (Fig 12). The artifact is barely detectable in the cold storm tops in this RGB and example, and the important features and colors can be analyzed as with GOES-16 and GOES-17.
The previous examples are from the GOES-18 5-min PACUS sector, but the “Barcode Artifact” appears similarly across the Full Disk Sector and Mesoscale sectors. The example below reveals the appearance of the artifact in mesoscale 1-min Nighttime Microphysics RGB imagery off the southwest US coast on the morning of June 9 (Fig 13).
Alaskan NWS users will also notice the “Barcode Artifact” in GOES-18 imagery. Given the longer cool season, longer nights in the winter, and cooler temperatures overall, the artifact should appear more often in this imagery. The examples in Figures 14, 15, and 16 show GOES-18 1-min grayscale Ch07 single band imagery (-60C to 100C), the Fog Difference (-15C to 15C), and Nighttime Microphysics RGB from 12Z to 15Z on the morning of June 30 over the Aleutian Islands. The artifact is present across the cool scene, especially in the high-level cloud tops, but the various cloud features are still diagnosed and differentiated. As alluded to previously, some areas of low clouds (blue) in cool air mass cases such as this one appear to flash in and out due to the dynamic pattern of the vertical striping.
From around Hawaii, the the Fog Difference is shared during an overnight period (Fig 17). The banding is much less obvious compared to over Alaska, given the relatively warmer scene involved.
This post shared GOES-18 examples of imagery products, available in NWS AWIPS, that include the 3.9 um channel as a component, in order to characterize the impact of the “Barcode Artifact”. While the artifact is not readily apparent in the single-band imagery, it becomes obvious in channel differences (such as the Night Fog Difference), and multispectral products (RGBs and Geocolor) that leverage the differences, as a series of shifting vertical striping. The artifact is most apparent in cold scenes, such as at night, and in relatively cold cloud tops. In most situations, analysis of a scene should not be significantly hampered by the noise, and the user will view it as a slight annoyance. In particularly cold scenes, and depending on the colormap and range used, a given cloud feature may appear to flicker. Engineers continue to work toward a solution to the GOES-18 Channel 07 “Barcode Artifact”.
This CIMSS Satellite Blog also shares information about the “Barcode Artifact”.
Bill Line, NESDIS/STAR and CIRA