A previous blog post documented the explosive growth of the East Troublesome Fire during the day of Oct 21 through the late evening. The fire had spread east to near the Continental Divide, west of Rocky Mountain National Park (RMNP), slowing by late that night.
During the morning of 22 Oct, NWS BOU forecasters monitoring the wildfire hot spot in GOES-East imagery noted an eastward movement of the far eastern portion of the fire, northeast of Grand Lake and west of Estes Park, possibly across the Continental Divide (Fig 1).
An image captured by BOU forecasters shows the Fire Radiative Power product in relation to the most recent burn scar shapefile and other local geographic features and towns (Fig 2). Accounting for the surface parallax (from GOES-East, surface features are displaced to the north and west at this location/elevation by several km), fire associated with the easternmost hot spot would actually be situated to the southeast, just east of the Divide and west of Bear Lake in RMNP.
Accounting for surface parallax, BOU believed that the hot spot may have advanced east across the Continental Divide during this period. Based on this development as diagnosed in GOES-East imagery, BOU forecasters alerted (via phone call) RMNP dispatch (and Laminar County) to the possibility that the fire had pushed east across the Divide into western RMNP. They were unaware of fire growth into the park at the time, and would go on to call out fire partners to investigate. Although it took a while to get confirmation, it would be confirmed that the fire had indeed crossed over the Continental Divide.
A cold front would soon push west into the I-25 corridor and eventually to Estes Park, dropping temperatures and raising humidity’s with a light east wind. The moist stable layer may have made it west up to the fire, putting a damper on fire behavior. GOES Natural Color Fire imagery from the early afternoon showed low stratus draped across the eastern Colorado plains, while the wildfire continued to burn hot west of the Divide in the presence of still dry and windy conditions (Fig 3). Also diagnosed in the imagery was a thick smoke plume with pyrocu spreading well east over the stratus deck. The smoke plume masked the hot spot in RMNP for the rest of the afternoon/evening.
During the evening of the 22nd, the glow associated with the fire in western RMNP could be diagnosed in (terrain corrected) VIIRS Day Night Band imagery (Fig 4).
This is a great example of a forecast office utilizing GOES imagery to provide potentially life saving IDSS to core partners.
Bill Line, NESDIS and CIRA (with input from NWS BOU)
The East Troublesome Fire, in Grand County, Colorado near Grand Lake and west of Rocky Mountain National Park, experienced substantial growth during the afternoon/evening of 21 October 2020. Dry environmental and fuel conditions, along with gusty winds, caused the fire to grow from 19,086 acres to 125,602 acres during the 1-day timeframe per Inciweb (see maps below).
GOES-East imagery captured the rapid growth of the associated hot spot signature. Throughout the event, NWS Boulder shared GOES-East imagery of the fire on social media to help inform the public of it’s evolution as it quickly spread east. A couple examples are shown below.
VIIRS imagery from the early afternoon captured the wildfire as it began it’s rapid growth (Fig 1). The Fire Radiative Power product provided a high resolution view of the heat associated the fire, highlighting a particularly active zone over the northeast portion of the fire (which would go on to continue to expand east rapidly). The underlay of VIIRS True Color imagery shows the associated smoke plume with pyrocu developing near the hot spot. This imagery is available online from the JSTAR mapper.
The daytime evolution of the wildfire is shown through the GOES-East Natural Color Fire RGB in Figure 2. The rapid growth of the wildfire begins after 2000 UTC, with the large smoke plume extending well east. Ashfall was abundant across downstream locations such as Fort Collins and Loveland. Pyrocumulus clouds were also abundant with the smoke plume.
A VIS-IR-SWIR combo animation extending after sunset highlights the development of the smoke plume, including eventual cooling of pyrocu to as cold as -60C after dark (Fig 3).
Long animations of GOES-East SWIR and Fire Temperature RGB show the full evolution of the wildfire hot spot growth on the 21st from Noon through around midnight (Fig 4-5). Steady growth/heating is observed through teh afternoon, before the rapid acceleration east after dark to near the Continental Divide. west of Rocky Mountain National Park.
Similar time periods but zoomed out images provide another perspective of the large growth and massive size of the fire (Fig 6-7).
The fire becomes so hot in areas that the signal in SWIR channel becomes saturated. This is a situation where the Fire Temperature RGB becomes a little more useful for those wishing to monitor fire heating trends the most active/hottest regions of the wildfire. Figure 8 from 0131 UTC compares SWIR with Fire Temperature RGB, exemplifying the power of the RGB to reveal more detailed temperature information after the SWIR channel becomes saturated. While the SWIR saturates, the Fire Temp RGB shows progressively hotter regions from red to yellow to white through it’s inclusion of the 2.2 um and 1.6 um bands, in addition to the SWIR.
GOES-West similarly displayed the evolution of the wildfire through the afternoon/evening (Fig 9).
A couple hours after midnight, SNPP and NOAA-20 VIIRS DNB NCC imagery revealed the glow of the now very large hot spot associated with the East Troublesome Fire, as well as the most active areas (Fig 10). The massive size can be compared with the City of Denver to the east.
The VIIRS Fire Radiative Power Product, shown earlier in this post, is also available at night, and shown in Figure 11. Again, the product provides a higher resolution view of the current location of the wildfire, along with the hottest areas.
Wildfires remained active across northern Colorado by 20 Oct 2020. The Cameron Peak Fire, west of Fort Collins, had grown to over 200,000 acres, the largest wildfire in Colorado recorded history.
Viewing GOES-East SWIR imagery over northern Colorado during the morning of Oct 20, a hot spot is barely apparent from the Cameron Peak Fire, just west of Fort Collins (Fig 1-2 top). From the SWIR and other channels, one easily finds that this is due to cloud cover. However, the western US has the benefit of overlapping 5-min (CONUS/PACUS) imagery from GOES-East and GOES-West satellites. Upon viewing GOES-West SWIR imagery, a hot spot associated with the Cameron Peak Fire is readily apparent through the morning (Fig 1-2 bottom).
Viewing Natural Color Fire RGB imagery, the quasi-stationary cloud masking the hot spot in GOES-East imagery is obviously situated to the east of the wildfire in GOES-West imagery, allowing for a clear view of the hot spot (Fig 3-4). This is a good visualization of parallax, and how clouds will appear situated at different locations relative to the surface in reality, and between GOES-East and GOES-West.
It is important for forecasters in the west to remember that they have two options for 5-min geostationary imagery, and that there are situations where one may provide additional insight over the other.
A shortwave trough brought strong, deep westerly winds to northern Colorado and an afternoon cold front on 14 October 2020. Analysis of GOES-East water vapor imagery reveals the shortwave dropping south through MT/WY and then east into the plains (Fig 1). The gusty winds and low RH along with continued dry fuels meant conditions were favorable for the rapid growth and spread of the Cameron Peak Wildfire, which had been burning for months in the mountains just west of Fort Collins.
The dangerous fire weather conditions did indeed cause the wildfire to grow considerably during the day, becoming the largest wildfire in Colorado recorded history at over 164,000 acres by early on the 15th, from 135,000 acres early on the 14th. The growth can be visualized in the fire information maps from Inciweb (Fig 2a). Evacuation zones expanded east to just west of Horsetooth Reservoir (Fig 2b).
Both GOES-East and GOES-West satellite imagery captured the evolution of the wildfire hot spot and smoke plume throughout the day. Given the satellite viewing angles and resulting forward scattering, GOES-West VIS provided a clearer view of the smoke plume during the morning, with GOES-East VIS the better option during the afternoon (Fig 3). Given the thick plume, both sensors provided adequate detection.
Focusing on GOES-East, 1-min imagery was available over the wildfire during the day. Early morning Natural Color Fire RGB imagery revealed a lenticular cloud stationary over the fire location around sunrise, dissipating into the morning and revealing the large hot spot (Fig 4).
The Natural Color Fire RGB imagery allows one to characterize various aspects of the wildfire given the three components: hot spot (SWIR), smoke plume (VIS), and burn scar (Veggie). While a similar RGB (Day Land Cloud Fires) is available in AWIPS, this particular RGB, which better detects hot spots, is not (though it can be requested). An example scene from this wildfire is annotated in Fig 5.
Later in the day, the wildfire broke containment and spread rapidly to the east. This expansion is shown in a 2.5 hour period of GOES-East 1-min imagery in the Natural Color Fire RGB (Fig 6).
The full daytime evolution of the wildfire in the Natural Color Fire RGB is shown in Figure 7. A similar animation is shown for the Fire Temperature RGB, which can be used to diagnose relative “hot” areas within the broader hot spot of a mature wildfire (Fig 8).
The smoke was present at the surface across Fort Collins from the morning through the early afternoon. However, a surface backdoor cold front pushed west into the I-25 corridor by mid-afternoon, clearing the near-surface smoke and dramatically improving air quality. The smoke plume remained aloft, however, as was shown in Fig 9, confirming the low-level nature of the cold front. IR-Window imagery with a grayscale color table captures the southwest evolution of the cold front and it’s minimal influence on the smoke plume aloft as observed by from satellite.
NOAA-20 VIIRS True Color Imagery and Active Fires Product around 2000 UTC (tail end of rapid spread east) provided a detailed view of both the smoke plume as well as the active fire burn area (Fig 10).
The natural color fire RGB can also be applied to VIIRS imagery (Fig 11). By using I-band imagery, the product becomes much more detailed given the 375 m spatial resolution. In this case, there were three VIIRS images available within a ~1.5 hour period from SNPP and NOAA-20, allowing for an analysis of the fire growth during that period. Note missing hot spot data within the larger hot spot due to band I4 (swir) pixel saturation. Land surface features such as wildfires are much easier to analyze in time in VIIRS imagery since the implementation of Terrain Correction for VIIRS EDR’s.
That evening, NOAA-20 VIIRS Day Night Band captured the glow associated with the Cameron Peak Fire, in addition to that from nearby city lights (Fig 13).
Some photography of the fire smoke plume from Oct 14 follows:
Very dry antecedent conditions and the passage of a cold front with strong post-frontal winds resulted in the development of a haboob and blowing dust across the southern plains during the evening of 11-12 Oct 2020. Images and videos from across the region captured the haboob as it progressed south and east across CO into KS/OK/TX during the evening.
One-minute GOES-East imagery captured the early evolution of the haboob along the cold front as it progressed south across southeast Colorado during the final hours of sunlight (Fig 1 and 2). The examples provide a comparison between a feature-relative and fixed region animations.
Given the presence of patchy cloud cover atop the blowing dust, RGB imagery could be used to more easily differentiate/confirm areas of blowing dust (Fig 3 and 4).
Ten-min GOES-East imagery covering a broader region and longer period over the southern plains captured the full evolution of the cold front/haboob and region of blowing dust. IR window imagery with a custom grayscale colortable and range of -55C to 45C clearly highlights the temperature contrast ahead and behind the cold front (Fig 5).
Dust-Fire RGB imagery highlights areas of blowing dust, wildfire hot spots, and intense smoke plumes in the strong southwest flow ahead of the front, in addition to the cold front itself (Fig 6).
The following morning, a bore was diagnosed in visible imagery following passage of the cold front across southeast Texas (Fig 7).
A long-lasting upper level ridge over the western US gave way to a relatively potent upper level trough on 07-08 Sep 2020, resulting in active weather across much of the western US. Over the Pacific Northwest, the system sent a cold front through the region resulting in very dry conditions with gusty winds during the day on the 7th. These conditions helped support the spread of large and fast moving wildfires, as well as widespread blowing dust emanating from freshly plowed fields. As a result, NWS Spokane, WA issued Wind Advisories, a Red Flag Warning and Blowing Dust Advisory for the area.
GOES-West 3.9 um shortwave IR imagery with a simple linear grayscale colortable captures the initial development and following rapid evolution of the wildfires well (Fig 1), while visible imagery reveals widespread opaqueness across the region (Fig 2). The visible texture and (warm) brightness temperature of the atmospheric aerosols (along with presence of wildfires) leads one to surmise that it is either smoke and/or lofted dust.
Combining the SWIR and VIS, it is revealed that some of the aerosols are anchored to hot spot locations, and are therefore likely smoke plumes, while others are originating from open fields with no hot spots, and are suspected regions of blowing dust (Fig 3).
When we view the SWD (with SWIR hotspots overlay), a reliable method for capturing lofted dust given sensitivity of the 10.3 um band, much of the opaque region (smoke and dust) provides a signal typical of lofted dust (neg 10-12 um diff; dark gray to black in this example; Fig 4). There is typically little-to-no signal for smoke in this difference.
As a result, the Fire Dust RGB, that combines IRW, SWIR, and SWD to capture hot spots and dust plumes, shows a similar signal between the lofted dust and active smoke plumes (Fig 5).
Viewing other wildfires in the west (Fig 6), there is a similar SWD signal for some of the most impressive smoke plumes that developed later in the day from the large/very active wildfires (Fig 7). Early day smoke across the area that is composed of much smaller particulates has a very weak to no signal in the SWD. The SWD signal apparent in the very active smoke plumes is likely associated with larger smoke particles (ash) being lofted high into the plume by the strong updraft generated by the wildfire. In the Washington case, the SWD signal is likely a brew of dust mixing with smoke and lofted ash.
Back to Washington, an alternate and IR-only RGB that replaces the IRW (from the Fire Dust RGB) with the Cloud Top Phase difference appears to do a slightly better job at differentiating lofted dust (cyan) from the intense/active smoke plumes (bright green) due to absorption differences between the two channels from dust (small and uniformly shaped particles) to smoke/ash (varying sized particles; Fig 8).
A zoomed out view of the same RGB over the whole western US during the day and following evening continues to separate the impressive smoke plumes from the blowing dust (Fig 9).
Combining the VIS, SWIR and 0.86 um veggie band into a Fire Day RGB discussed in previous blog posts, the lofted smoke and dust become more obvious, and one can diagnose a slight difference between the most probable dust regions (greener cyan) and smoke plumes (bluer cyan), in addition to the hot spots (Fig 10). NWS Blowing Dust Warning polygons are overlaid on the imagery.
GOES-West Geocolor Imagery also captures the smoke and dust well, with slight differences between the two aerosols discernible (Fig 11). GOES-East Geocolor also captures the plumes, particularly later in the day as forward scattering increases toward that satellite (Fig 12).
SNPP and NOAA-20 VIIRS Day Night Band NCC imagery captured the glow of the wildfires across Washington (Fig 13). The first few images in the animation are from the 6th, and show the scene (day and night) prior to fire ignition. During the overnight hours early on the 7th, the first large fire developed and was apparent in the series of VIIRS passes. The following day, the initial fire grows and others ignite, with smoke obvious in the imagery. During the overnight hours early on the 8th, the wildfires had grown considerably, and were depicted in the VIIRS DNB imagery. In particular, the perimeter of the wildfires, along with the most active areas, are captured well in the DNB imagery.
Many images and video depicting the degree of visibility reduction by dust and smoke were shared on social media, some of which are included below.
Hurricane Laura became a named tropical system in the Caribbean at 1500 UTC 21 Aug 2020, and a Hurricane at 1500 UTC 25 August 2020. According to the NHC very early on 27 Aug, “Laura made landfall near Cameron, Louisiana, around 0600 UTC (1 am CDT) with maximum sustained winds of 130 kt, which is near the high end of category 4 status.” The following post includes a collection of GOES ABI imagery captured during the evolution of Laura.
The full evolution of Laura as a named storm through the day after landfall (21-27 Aug) is shown in Figures 1-3 as hourly GOES-East animations. Figure 1 includes 10.3 um IR window channel imagery, while Figure 2 transitions between 10.3 um IR window channel imagery during the night, and VIS-IR Sandwich imagery during the day. Figure 3 characterizes lightning activity during the life of the storm, utilizing GOES-East Flash Extent Density (note, only 5-min GLM FED was used).
A water vapor animation with RAP 500 mb wind and height analyses captures the influencing large scale features during the long trek of Laura. Notably, a broad ridge over the western Atlantic early in the period expands west into the southeast US and eastern GoM throughout the animation, helping to steer Laura west of the track of the preceding Marco, into the western GoM (Fig 4).
A feature relative GOES-East VIS animation during the full day of the 25th depicts the strengthening of Laura from a Tropical Storm to a Hurricane (Fig 5).
Zooming out for the same period, Laura is seen advancing into the central Gulf of Mexico, while remnants of Marco accelerates west along the Louisiana Gulf Coast (Fig 6).
A mesoscale sector was available over Laura during it’s evolution, providing forecasters with valuable 1-min-updating imagery. The final 70 minutes of visible imagery on the 25th capture increasing thunderstorm activity around the center of circulation (Fig 7). One-minute imagery eases diagnosis of a center of circulation in tropical systems, particularly in unorganized storm systems. The evolution of individual convective updrafts associated with the tropical system are also more efficiently tracked in space and time using the high temporal resolution, low latency imagery.
During the overnight hours of the 25h-26th, Laura continued to strengthen, with an eye becoming apparent by the early morning of the 26th per GOES-East IR imagery (Fig 8).
Sunrise over Laura on the 26th revealed a much better organized hurricane with a developing eye, albeit still contaminated with some cloud debris (Fig 9).
A zoomed out view of the full mesoscale sector shows the massive storm approaching the coast (Fig 10). The IR-VIS sandwich combo imagery combines the high spatial detail of the VIS with the quantitative BT information from the IR.
By the late morning of the 26th, the eye had cleared considerably, and low and upper level vorticies could be diagnosed in the 1-min VIS with convective activity still becoming organized within the eyewall (Fig 11).
During the afternoon, eye clearing had completed, convective activity became more consistent within the eyewall, and a healthy major hurricane was apparent (Fig 12).
A clear eye and healthy eyewall were still apparent in 1-min visible imagery as sun set on the storm during the early evening of the 26th, jsut several hours prior to landfall (Fig 13).
The full development of the impressive storm during the day of the 26th is diagnosed in GOES-East visible imagery (Fig 14).
GOES-West provided a unique perspective of the hurricane on the 26th given the much larger viewing zenith angle compared to that of GOES-East (Fig 15).
Landfall of Hurricane Laura in southwest Louisiana was displayed in GOES-East IR imagery during the overnight hours. Imagery shows the large eye remaining intact well inland, before filling in by early morning (Fig 16).
Figure 17 provides a zoomed in look at 2-min IR imagery during landfall, including surface obs.
GOES-East visible imagery after sunrise on the 27th shows the massive storm and lack of clear eye (Fig 18). The weakening tropical system filled most of the 1000 x 1000 km mesoscale sector.
Finally, Day Cloud Phase Distinction RGB imagery from the 27th shows convective activity and upper level clouds (reds and yellows) becoming detached from the low level circulation (cyan/blue clouds; Fig 19).
A long-lived line of severe thunderstorms resulted in a broad swath of damaging winds across the Midwest on 10 August 2020. There were hundreds of severe wind reports associated with this derecho, including dozens in excess of 75 mph (significant severe). GOES-East captured the evolution of the complex from initial thunderstorm development over Nebraska through convective decay over Ohio.
A long, 10-min IR animation captured the full evolution of the thunderstorm complex, from 0611 UTC with initial development of thunderstorms over Nebraska, through 0201 UTC with weakening over Ohio/Indiana (Fig 1). Persistent cold cloud tops of <60C were analyzed along the leading edge of the MCS and in association with the severe storms, with cloud tops as cold as -80C sampled in the GOES imagery. NWS convective warning polygons and Local Storm Reports are shown as an overlay on the imagery.
Corresponding GLM Flash Extent Density imagery is shown in Figure 2, and is used to infer the locations of strongest updrafts, and updraft trends, within the broader complex. Periodic long flashes are also observed extending into the thunderstorm anvils, representing a lightning threat well away from the strong thunderstorms.
Corresponding MRMS composite reflectivity is shown in Figure 3 for comparison, and aligns with the regions of notable/persistent GLM activity and coldest cloud tops.
GOES-East VIS-IR Sandwich image combo (every 5-minutes) is shown as feature following zoom for the during the daytime of the 10th, following the derecho (Fig 4). The evolution of features within the thunderstorm line is made more apparent in the feature relative animation. The combination of texture in the VIS and brightness temperature information in the IR allows for easy diagnosis cloud top health and trends, including that of overshooting tops, gravity waves, overall texture, and above anvil cirrus plumes. Toward the end of the animation, cloud tops begin to warm, and texture becomes less abundant, indicating weakening convection.
GOES-East mesoscale sectors were available over the region, providing 1-min imagery for forecasters (Fig 5 and 6). The high temporal resolution, low latency imagery allows forecasters to more effectively track individual updraft trends in real-time vs the 5-min imagery.
A long (5-hour) 1-min VIS-GLMFED Sandwich animation covers a period of some of the most intense thunderstorm wind gusts, and connects visual texture trends with lightning trends (Fig 7).
On 28 July 2020, a new “Terrain Correction” was applied to SNPP and NOAA-20 VIIRS Imagery EDR geolocation thanks to work done by the VIIRS EDR Imagery Team. The terrain correction software provides consistent navigation of a given surface pixel, no matter the elevation or position within a swath. Prior to the change, high elevation pixels would appear to shift location within a scene from swath to swath as a result of their changing position within the swath relative to nadir. Examples of the change are shown below. The message from NESDIS:
First we analyze a scene over WA/OR, with the high elevation Cascade Mountain Range flanked by lower elevations to the west and east (Fig 1). Two daytime swaths of NOAA-20 VIIRS contained the scene on 27 July 2020, at 1937 UTC (western part of swath) and 2119 UTC (eastern part of swath). I1 band (0.64 um VIS) EDR imagery appears to depict a shift in the mountain range from west to east from the 1937 UTC swath to the 2119 UTC swath, while the position of low elevation areas within the scene remain relatively static.
Now viewing the same scene/imagery but on 31 July 2020, with terrain correction applied, there is very little (if any) shift in terrain (Fig 2).
A similar daytime example is shown using SNPP VIIRS I1 band imagery over south-central Alaska (Fig 3). On 14 July 2020, the position of the mountains within the scene appear to shift dramatically from 2119 UTC to 2301, while the adjacent lower elevations experience no shift at all.
The same scene on 04 August, following the terrain correction, experiences very minimal shift of the mountains from swath to swath (Fig 4).
Another example is applied to the VIIRS Day Night Band Near Constant Contrast EDR product at night (Fig 5). The first example, from 16 June 2020, is centered over northern Arizona and the active Mangum wildfire. Much of the scene is at an elevation between 4500 ft and 8000 ft, with the wildfire around 7500 ft. From 0837 UTC to 1014 UTC, illumination associated with the wildfire, and nearby towns, appear to shift from west to east.
Now viewing a similar scene over western Colorado from the late night of 04 August, after the terrain correction (Fig 6). The scene also contains a wildfire, and similar elevation range as previous. As we compare swaths, however, the light associated with the high elevation wildfire and nearby towns remain stationary.
Georges Bank is a region relatively shallow waters oriented SSW to NNE in the western Atlantic Ocean just east of Cape Cod (Fig 1).
An upward flux of relatively cool subsurface water onto Georges Bank due to tidal mixing (high tide and low tide due to the moon) yields cool Sea Surface Temperatures (SST, relative to surrounding waters) during summer months. The process also results in a nutrient-rich aquatic environment, resulting in a rewarding fishing location for New England.
Given the relatively cool SSTs over Georges Bank during the summertime, northward advection of moist low-level air over the geographic feature results in the development of widespread low clouds and fog. This, of course, poses a hazard to marine interests, particularly fisherman. The NWS (BOX) is responsible for issuing forecasts to only 40 miles offshore (coastal waters forecast), but fog and low clouds do impact those waters too, so forecasters will mention fog conditions when expected/observed. They also receive and respond to inquiries from mariners regarding offshore conditions. Satellite imagery is an important observational tool in providing these forecast products and decision support services.
Low clouds and fog developed over Georges Bank during the day and evening of 28 July 2020. GOES-East SST derived product highlights the cool temperatures over Georges Bank, surrounded by warm waters to the north and south (Fig 2). SSTs sampled over the bank during the morning of the 28th were in the mid 60s F, with low-mid 70s to the north, and mid-upper 70s to the south.
Animating the hourly SST product with RAP SFC wind barbs and dew point temperatures overlaid, moist southwesterly flow was analyzed with upper 60s to low 70s dew point temperatures entrenched over the relatively cool, mid 60s, waters of Georges Bank (Fig 3).
The result of a warm, moist low-level airmass progressing over the cool waters of Georges Bank on 28 July was the development of low clouds and fog. GOES-East 10-min visible imagery reveals the development of clouds over Georges Bank, and advection to the northeast, during the daytime hours (Fig 4).
GOES-East Day Cloud Phase Distinction RGB (modified to enhance low cloud appearance: Green max to 40, blue max to 30) similarly shows the development and evolution, confirming low-liquid water clouds, or bright cyan colors, and differentiating the low clouds from other features (Fig 5).
Overnight, the low clouds continue to be detected using typical IR band combinations. The Night Fog Difference (10.3 – 3.9 um) highlights the low clouds as positive values (blue) due to a difference in emissivity from low clouds at the two wavelengths, contrasting well with the near-zero difference (light gray) from the background surface and upper (ice) clouds in the foreground (Fig 6).
Combining the Night Fog Difference with the IR-Window Channel (13) and Split Window Difference (12.3 – 10.3 um) results in the Nighttime Microphysics RGB, which provides more inclusive cloud detection at night (Fig 7). The evolution of the low (water) clouds and fog is captured in the RGB as aqua/light blue, along with other cloud layers such as mid-level (liquid) clouds (light green) and high thick (red) and high thin (black/dark blue) ice clouds.
There is a drawback to these nighttime IR detection methods in this region. Viewing the Fog Diff over a period surrounding sunset, one notices relatively static regions of constantly positive values (blue) prior to and following sunset in the cold water locations under clear skies (confirmed in visible imagery), especially along the Maine north coast and east around Nova Scotia in this example (Fig 8).
These static regions of consistently positive fog difference values during the day and night are not clouds, but are due to water vapor sensitivity differences between the 10.3 um and 3.9 um bands, abundance of water vapor in the low levels, and the presence of a low-level temperature inversion (which is especially strong over the cold water zones). Compared to at 3.9 um, moisture absorbs (and re-emits) energy better at 10.3 um. Therefore, the satellite is sensing a layer at a higher altitude at 10.3 um compared to at 3.9 um (especially in a very moist environment). Given the temperature inversion, the higher altitude environment sensed at 10.3 um is warmer than that closer to the surface at 3.9 um. As a result, taking the 10.3 um minus 3.9 um difference will yield a positive value, similar to that for liquid clouds, making it difficult to confidently assess the presence of low clouds and fog in these conditions.
A half-hourly animation of the Geocolor product from early on the 28th through the overnight hours shows the complete evolution of this low cloud event (Fig 9). Overnight, when the product utilizes the fog difference for highlighting low clouds (blue), the influence of the aforementioned false alarm on the Geocolor product are readily apparent. One must keep this phenomenon in mind when analyzing low clouds and fog at night in RGBs and other products utilizing the fog difference, particularly when a low-level inversion and moist layer is present.
Moving ahead to Aug 2, we again analyze a geocolor animation from overnight through the early daytime (Fig 10). The low cloud false alarm “blue” areas are present overnight, and are confirmed after sunrise when the cloudy areas are obvious vs the clear sky ocean.
A VIIRS Day Night Band Near Constant Contrast image during the overnight hours confirms the location of low clouds over Georges Bank and the eastern part of the scene vs false alarm around Nova Scotia and portions of Georges Bank (Fig 11).
Bill Line (NESDIS/STAR), Louie Grasso (CIRA), Eleanor Vallier-Talbot (retired NWS BOS)