Satellite Liaison Blog

Building off of a previous blog post, a simple RGB can be made that allows for the observation of the hot spot, smoke plume, and burn scar associated with a wildfire. The RGB discussed in this post combines the 3.9 um band (RED) to sense the hot spot, the 0.86 um band (GREEN) to highlight the previously burned area, and the 0.64 um band (BLUE) to track the smoke plume. The hot spot (active wildfire) will appear as red, the smoke plume as faded blue or cyan, clouds a bright cyan, and burned area as a locally dark area. Highly vegetated areas will appear as a bright green, and bodies of water very dark. The recipe used in this example is shown in Figure 1. An animation of this RGB with GOES-17 for the Kincade Fire on 27 Oct is found in Figure 2, and the RGB with the three ingredients is shown as a 4-panel in Figure 3. The active large fire is readily apparent, with the associated burn scar extending north of the ongoing fire. The smoke plume is diagnosed extending well to the southwest of the fire. The heavily forested region of northern California is obvious to the west and northwest of the fire. What appears to be lofted dust is also apparent in this example in the southeast part of the scene.

Bill Line, NWS

A late October trough brought significant weather impacts to portions of the western and central United States. The impressive trough was diagnosed in GOES-East water vapor imagery with features readily apparent (Fig 1). The upper jet extended south across the Pacific Northwest, rounded the base of the trough, and stretched northeast across the Great Basin and into the northern US plains. Exceptional shortwave energy near the base of the trough was digging south across northern California, and an associated cold front was racing south down the southern high plains. An overlay of GOES-16 Derived Motion Winds (DMWs) confirms the speed of the jet in areas where winds are available.

In California, extreme fire danger was observed as high pressure built in the wake of the potent shortwave, resulting in strong easterly surface winds and plummeting RH in the presence of very dry fuels. In northern California, rapid expansion and increase in temperature of the Kincade Fire was observed by GOES-West shortwave IR imagery during the overnight hours of the 26th into the early morning hours of the 27th (Fig 2).

Further east, the precise location of the cold front could be tracked as it pushed south across the high plains, with cold air pooling along the Colorado front range and southeast mountains (Fig 3). An overlay of MSAS 3-hr pressure changes and wind barbs shows an expected increasing pressure behind the IR-detected front, along with a shift of winds to a northerly direction.

With sunrise, the extensive shield of low clouds developing within the cold airmass in the wake of the cold front was observed in GOES-East visible satellite imagery (Fig 4). Pikes Peak and other mountain ranges are seen poking through the low cloud layer. Additionally, a multitude of cloud top gravity was are diagnosed atop the cloud layer.

Bill Line, NWS

A cloud-to-ground lightning strike initiated a wildfire in southeast Colorado during the early evening of 18 October 2019. A progressive shortwave trough moving across Colorado sent a cold front south down the eastern Colorado plains during the afternoon and evening. Lift associated with the trough and cold front, and weak instability aloft, aided the development of showers and weak thunderstorms ahead of and along the frontal boundary. Dry low levels and dry fuels in place across the eastern Colorado plains along with gusty north winds behind the front aided the growth and southward spread of the lightning-initiated wildfire during the evening. Given the expected conditions, a Red Flag Warning for dry lightning was issued the morning of the 18th.

GOES-East water vapor imagery shows the shortwave trough quickly moving east across Colorado on 18 Oct (Fig 1). The associated frontal boundary is also diagnosed in water vapor imagery racing south across the central high plains.

The National Lightning Detection Network (NLDN) detected a negative cloud-to-ground lightning strike over SSE of Lamar at 2315 UTC (Fig 2). Interestingly, neither the GOES-West nor GOES-East GLM detected the flash associated with this particular CG. Shortly after the CG detection, under clearing cloud cover, a hot spot was noted in GOES-East 3.9 um shortwave IR imagery at the location of the CG strike. The fire accelerated south thereafter as north winds picked up behind the front. NWS Pueblo provided spot forecast information to local fire crews working the wildfire overnight.

Overnight, the JPSS NOAA-20 VIIRS Day Night Band detected light associated with the wildfire during the late night hours, along with the scorched earth in its wake to the north (Fig 3).

The next morning, GOES-East visible imagery showed the north-to-south oriented path of scorched earth associated with the wildfire emanating south from the location of the previous day’s lightning strike (Fig 4).

Bill Line, NWS

The Decker Wildfire has been burning just a few miles south of Salida, CO in the far northern Sangre de Cristo wilderness since 8 September 2019. As of 13 October 2019, the fire had burned 8,118 acres and has prompted periodic evacuations and pre-evacuations. On 13 October 2019, the fire had broken containment during critical fire weather conditions. The intensification could be seen in GOES-East 3.9 um SWIR imagery via the flare up in brightness temperature south of Salida around 18Z (Fig 1).

The smoke plume was easily diagnosed in GOES-East visible imagery extending well east of the fire within strong westerly flow (Fig 2). A significant increase in smoke production was observed after 18Z, following the flare up seen in the SWIR imagery.

SNPP VIIRS True Color imagery with VIIRS Active Fires product overlaid during the early afternoon shows numerous thermal anomalies (~750 m spatial resolution) associated with the fire along with the extensive smoke plume (Fig 3).

The IMET tasked to the fire requested that WFO PUB request a mesoscale sector in support of the fire fighting activities. WFO PUB requested another mesoscale sector the following day (10/14) given continued critical fire weather conditions over the fire.

A photo taken around 2300 UTC from between Canon City and Pueblo shows the impressive smoke plume around sunset (Fig 4).

A S-NPP pass during the night of the 13th provided VIIRS Day Night Band imagery over Colorado with favorable illumination. The Decker Fire is readily apparent in the imagery as a cluster of bright light south of Salida in a region that would otherwise be dark.

Bill Line, NWS

An early season winter storm brought much colder temperatures and widespread snowfall to portions of the eastern Rockies and high plains October 9-10. Analysis of GOES-16 water vapor imagery shows key large scale features associated with the system as it digs south into the Great Basin and WY/CO (Figure 1) through 12Z. Overlaid on the animation are 700-300 mb GFS-derived Quasi-Geostrophic Omega, highlighting regions of greatest ascent and descent, correlating with what is observed in the imagery. The surface cold front is also seen pushing south through the high plains.

Now analyzing GOES-16 IR imagery over the same period and zooming in, the southward progression of the cold front is easily diagnosed in the imagery, including the banking of cold air up against the Colorado front range and Sangre de Cristo Mountains. An overlay of a surface wind analysis confirms the progression of the front. 60 mph winds were measured behind the front across southern Colorado.

Behind the front in northern Colorado during the evening of the 9th, thunderstorms managed to develop, producing heavy graupel and small hail. GLM Flash Extent Density from GOES-16 showed a lightning jump during the development of the strongest storm, which produced up to 3/4″ hail.

The Suomi NPP VIIRS Day Night Band (DNB) Near Constant Contrast (NCC) product provided high resolution “visible” imagery during the night as the front pushed south and low clouds expanded across the plains, thanks to illumination from the moon.

The storm system resulted in widespread snowfall amounts of up to around 5 inches over the portions of the Colorado I-25 corridor and eastern plains. The Day Cloud Phase Distinction RGB can be utilized during winter weather events to diagnose developing bands of snow or track ongoing snow bands during (especially in poor radar coverage areas). Using this event over southeast Colorado as an example, shades of cyan colored clouds (water clouds) transitioning to bright green represent increase of ice in the cloud top and a potential snow producing cloud/band (Fig 5). Multiple snow bands developed in the vicinity of Pueblo and areas south, expanding east through the afternoon in the presence of strong upper forcing and low-level easterly/northeasterly upslope flow.

Overlaying base radar reflectivity, one can see the snow bands as observed in radar imagery match up well with our analysis of the bands in RGB imagery (Fig 6).

NOAA-20 VIIRS DNB NCC imagery from the next evening provided an early view of snow cover over eastern Colorado from the previous day’s storm (Fig 7). Widespread snow cover is observed along the I-25 Corridor from Colorado Springs to Fort Collins, and much of the plains to the east. Further south, snow cover resulting from the banded snowfall is diagnosed near and south of Pueblo. Low clouds are masked with the GOES fog difference (blue).

Bill Line, NWS

NWS forecasters have the ability to make modifications to the three RGB components (Red, Green, Blue) in AWIPS. These modifications can be made easily by right-clicking (and hold) on the loaded RGB in the product legend, and selecting the “Composite Options” (Fig 1). Slider bars and numerical values for the max and min values, along with the Gamma, for each of the Red, Green, and Blue composition of the RGB will appear. The forecaster can adjust the numbers using the sliders or by editing the numbers, which will adjust the appearance of the RGB on the fly. It is recommended these adjustments be made on the fly as opposed to saving a procedure as desired thresholds will change depending on the time of day, season, and situation. The goal of the adjustments is to enhance the usefulness of the RGB by drawing out relevant features in certain situations.

There are occasions where making these modifications will extend or improve the usefulness of the RGB, particularly for RGBs which utilize VIS or NIR channels, and therefore depend on sunlight. The Day Cloud Phase Distinction (DCPD) RGB is useful for tracking cumulus cloud evolution from early water cloud (cyan), to convective initiation and glaciation (bright green), to mature convection (yellow). Because this process often takes place midday during the summer, the relatively low max thresholds of the reflectance/albedo components (79% for the 0.64 um vis green component, 59% for the 1.61 um snow/ice band blue component), are often exceeded, causing those components to wash out and provide no texture detail. Therefore, it is beneficial to increase the thresholds for those VIS/NIR fields to ensure saturation of the field will not occur. The exact thresholds to set will depend on the situation, including time of day, time of year, location, etc. These changes should only take around 15 seconds to make, and may need to be updated periodically during the day as the sun angle changes. A video from the 2019 Satellite Applications Workshop exemplifies the process of making adjustments to the DCPD RGB is such a situation.

A convective initiation example from 9 August 2019 over Colorado compares the DCPD RGB before (Fig 2) and after (Fig 3) a change to the RGB recipe was made. The forecaster was monitoring the scene for signs of imminent convective initiation, and made the changes to the RGB on the fly. A comparison of the values for the default RGB and modified version, along with the corresponding imagery at 2014Z, is included in Figure 4.

It is also beneficial to modify the DCPD RGB during low light situations, just after sunrise and before sunset, in an effort to extend its usefulness. During these times of day, features will be more difficult to discern due to low albedo from the green (0.64 um) and blue (1.6 um) components relative to the ranges set in the RGB. Therefore, adjusting the upper threshold downward for both will make cloud features apparent slightly earlier in the morning and later in the evening. If doing this in the morning, the user will need to be sure to return thresholds upward as albedo increases.

A common situation where such low light adjustments are helpful is for low cloud monitoring around sunrise or sunset. For example, early morning low clouds in the San Luis Valley of CO/NM on 12 August 2019 were difficult to visualize in the RGB with default settings (Fig 5). Adjusting the upper bounds downward for the VIS and NIR components, the fog (aqua colors) becomes brighter and easier to diagnose (Fig 6). The adjustments and direct comparison are shown in Figure 7.

Alternatively during low light situations, the Gamma values for the reflectance components in an RGB can be lowered in an effort to put more weight on them. An example of valley fog over PA/NY compares decreasing the Max values with decreasing the Gamma values for the VIS (green) and NIR (blue) components in the DCPD RGB (Fig 8). Both adjustment strategies similarly make the fog more apparent when compared to the default settings. Some combination of Max/Gamma reduction would also work to draw out the low level cloud features.

The VIS/IR Sandwich RGB similarly benefits from adjustments during low-light situations. Lowering the max threshold and gamma for each of the three components equally results in a brighter picture with clearer detail in convective cloud tops. An example of the changes made for a likely severe storm in southern Colorado during the evening of 13 September 2019 is shown in Figure 9. Features such as cumulus clouds, overshooting tops, and above anvil cirrus plumes are more easily and quickly detected in the modified/brighter imagery.

Forecasters across the northern US plains have found utility in the Day Snow-Fog RGB for identifying and tracking areas of blowing snow. This RGB includes two VIS/NIR components, and therefore its use can be extended further into the morning and evening by making simple adjustments (Fig 10). Carl Jones (FGF) and Andrew Ansorge (DMX) discussed this task at the 2019 Satellite Applications Workshop, and specifically mention the value of making on-the-fly adjustments to the RGB in an effort to make features clearer during low light situations.

Alternatively, there are also situations where these modifications may change the meaning of the colors in the RGB, and could lead to misinterpretation of important features. Additionally, making adjustments to an RGB and then sharing it with others who are not aware of the specific adjustments and interpretation of the new scheme can be dangerous. Therefore, forecasters making adjustments should be sure to understand how their changes are influencing the meaning of the final RGB, and should probably not save and share modified AWIPS RGBs with others. Rather, the process or practice of modifying an RGB to help in specific situations should be shared.

Bill Line, NWS

An isolated and long-lived thunderstorm developed along the Raton Mesa during the late afternoon of 13 September 2019 in an environment characterized by weak forcing but favorable SBCAPE (~1500 j/kg) and EBSHEAR (~35 knots). While probability was low that a storm would develop, any storm that did develop had the potential to be strong to severe. Surface southerly to southeasterly upslope flow onto the Raton Mesa and low level convergence with the deepening surface lee trough would be the drivers for initiation. Early day HRRR runs hinted at the potential for very isolated storm development in this area as well.

Clear skies prior to cumulus cloud field development and deep convective initiation (CI) allowed for the analysis of the split window difference early in the day. There were no obvious signs of moisture pooling or boundaries in the area of future CI. However, upon crunching a linear grayscale colortable, a subtle moisture maximum corridor is apparent along the northern edge of the Raton Mesa between KTAD and K4V1, east of KVTP. This would imply relatively high amounts of low level moisture, maybe low-level moisture convergence/pooling, and an opportunity area for CI. Additionally, there appears to be a surge eastward out of La Veta Pass (KVTP) prior to CI, possibly enhancing low level moisture convergence up the northern portion of the Raton Mesa. These signs, along with what the HRRR had been showing, should give a forecaster increasing confidence that this particular area probably has the best chance for CI.

Convergence along the leading edge of this surge is apparent in radar imagery progressing toward the moisture maximum just prior to CI.

GOES-16 5-min Sandwich RGB imagery provided some helpful insight during the evolution of this event. During the early stages, clumping cu were becoming more apparent northwest of KTAD after 20Z, with an orphan anvil developing around 21Z off of the easternmost towering cu (fig 3). This feature indicates a failed convective initiation attempt and that a capping inversion is still present but may be on the verge of breaking. Convection successfully initiated from the same area shortly thereafter.

After convective initiation, the IR component to the sandwich appears and rapid cooling is apparent. Pretty soon after CI, an overshooting top is diagnosed, and an associated above anvil cirrus plume shortly thereafter. It appears as though plume generation likely began between 2220Z and 2230Z, at least 15 minutes prior to Maximum Estimated Size of Hail (MESH) first exceeding 1″. One-minute imagery would have made detection and tracking of this feature easier. Figure 4 zooms in on the area of interest and ends on a time when both the OT and plume are obvious in the sandwich imagery.

GOES-16 sandwich RGB imagery for the full duration of the storm through sunset and with no annotations is included in Figure 5. The storm remained over rural areas so warnings were not verified. MESH algorithm indicated hail larger than 1″ in diameter for much of the time between 2245Z and 0055Z, with a peak over 2″ around 0030Z (end of the animation).

Bill Line, NWS