In this FDTD GOES Applications Webinar, the Huntsville NWS office discussed the history of total lightning data usage at the HUN NWS Office, and described the usage of GLM data during a couple of recent severe weather cases. On December 16, 2019, the GLM data were used to diagnose severe convection in a moderately unstable environment. On January 11, 2020, GLM data factored heavily into warning decisions for severe convection, including tornado warnings, when their primary radar (KHTX) failed.
Below is an example of how GLM data were used in messaging during one of the events:
“Once the storms entered Marshall county, we really started to see an uptick in lightning activity, the strongest that we had seen that day. So we got on the radio and social media, and said, ‘we are really confident this storm is rapidly intensifying, something can happen here, … people in the path really need to be taking shelter right now.'”
FDTD Satellite Applications Webinars are peer-to-peer learning; staff from WFOs, National Centers, CWSUs, RFCs lead the presentations. The presentations are short (less than 30 minutes) and recorded for on-demand viewing.
Ideal conditions for the development of heavy freezing spray developed across the coastal waters of the northern Gulf of Alaska, including Cook inlet, on the back edge of a low pressure system during the afternoon/evening of 10 Feb 2020. Analysis of GOES-West water vapor imagery reveals the associated upper level trough exiting to the east as the next system approaches from the west (Fig 1).
Given the expected surface conditions, NWS Anchorage issued a Heavy Freezing Spray Warning for the associated offshore waters zone (Fig 2).
During the late afternoon and early evening of the 10th, surface observations indicated the development of gusty winds, temperatures well below freezing, and rough seas; all conditions necessary for the development of freezing (salt water) spray. Gusty winds and temperatures well below freezing were reported at Stations AUGA2 and AMAA2 (Fig 3), with wave heights of 10-15 ft reported at 46080. This is the region within which heavy freezing spray was expected and likely occurred based on analysis of VIIRS and ABI satellite imagery.
Given the relatively high latitude of the region, three VIIRS passes (one from NOAA-20, two from SNPP) were available during the day within 2.5 hours of each other. The five 375 m I bands from SNPP VIIRS for 2123 UTC are shown in Fig 4, centered over the region of freezing spray. A modified gray scale color table was created to focus on the reflectance values of the spray. The higher spatial resolution of the VIIRS imagery (vs GOES) captured the phenomenon in enhanced detail, allowing for easier diagnosis as to where freezing spray was occurring at that moment. The spray is observed as a region of relatively high reflectance (lighter gray) vs lower reflectance open sea extending from station AUGA2 through Cook Inlet and station AMAA2 into western portions of the broader Gulf of Alaska. Viewing VIIRS 375 m channels I1 – I3, it is obvious that the spray is most apparent in channel I2 given the relatively high reflectance of the lofted sea particles over the very low reflectance ocean surface.
Now viewing I4 and I5, the spray has a higher brightness temperature (darker gray) in the I4 (3.9 um) channel vs I5 (11.45 um) channel as a result of added reflectance component during the daytime due to scattering of the airborne particles. The brightness temperatures are similar in areas of clear sky with no spray over the ocean. Taking the difference between the two channels provides a clear view of where the spray is occurring. Sea current patterns are also apparent in bands I4 and I5, and are differentiated from the sea spray, particularly in band I4 where the spray has a higher brightness temperature, and currents have a lower brightness temperature.
Figure 4: 2123 UTC 10 Feb 2020 SNPP VIIRS I bands 1-5 and band 4 minus 5 difference (left to right, top to bottom). Higher res: I1 (0.64 um), I2 (0.865 um), I3 (1.61 um), I4(3.74 um), I5 (11.45 um), I4 – I5. Color table: low reflectance and warmer brightness temperatures is dark gray, high reflectance and cooler brightness temperatures is light gray. For the I4-I5 difference, dark gray represents a greater positive difference, light gray is near 0 difference.
Imagery from the three daytime VIIRS passes (2038 UTC from NOAA-20, 2123 UTC and 2306 UTC from SNPP) provides a sense of evolution of the spray during the day at high spatial resolution (Fig 5). Clouds appear to have developed within the region of spray by 2306 UTC.
GOES-West Full Disk sector provided high temporal resolution imagery over the region of lofted sea particles. At such a high latitude, the imagery spatial resolution is degraded (pixel area increases by roughly 4x), but still useful for detecting the spray in a broad sense. Given our analysis of the VIIRS imagery, we take a look at the 1 km (at nadir) 0.86 um band from GOES-West, and are able to track the evolution of the sea spray through the afternoon (Fig 6). A similar color table is used to that presented with the VIIRS imagery.
The CIRA Snow/Cloud product was very effective at capturing the spatial extent of the sea spray without needing to make any modifications (Fig 7). The integrity of the RGB for tracking other features is, therefore, maintained.
The ability to detect and track sea spray could be useful to NWS forecasters in verifying forecast products such as a Heavy Freezing Spray Warning, and for issuing new forecast products.
A shortwave trough digging into northern Mexico on 5 Feb 2020 brought gusty winds to the surface, leading to areas of lofted blowing dust, primarily from sources marked at points A and B (Fig 1). The lofted dust had traveled as far as Houston, TX, per media reports of dust being deposited at the surface.
There were two regions in particular from which a significant amount of surface material was lofted and subsequently carried a long distance. These locations, marked in Fig 2 and 3, include red earth from central Zacatecas (point A), and sand from southwest Coahuila (point B), where many sand dunes are present.
GOES-East 5-min CONUS imagery captured the onset of lofted dust from the aforementioned regions, along with it’s evolution as it wrapped around the southeast portion of the trough and was carried northeast into south Texas. The high spatial resolution afforded by the 500 m 0.64 um visible band provides the most detailed look of the lofted dust as it elaves it’s source, particularly toward sunset (Fig 4). Figure 5 provides a zoomed in look at lofted dust from the red earth (point A) region.
The Geocolor product developed at CIRA combines Channels 1, 2, and 3 and additional computations (making up for the lack of a green channel) to create pseudo-true color imagery during the day. In this case, the daytime geocolor imagery captured the dust plumes quite well, and differentiated the lofted red earth (shades of red) from the red earth region in the south and lofted sand (tan/gray) from the sand dune region to the north (Fig 6). The two source regions also appear red and tan, respectively.
Infrared imagery can also be used to capture lofted dust. The 10.3 – 12.3 um split window difference, previously discussed here for detecting dust, provides a very clear dust signature (negative difference values, dark gray to black; Fig 7).
Including the split window difference in the Dust RGB, along with the SO2 difference and 10.3 um IR window band, allows for dust detection (bright magenta to pink) along with cloud classification (Fig 8).
A daytime SNPP VIIRS pass at 1926 UTC over the area provided high resolution still images of the lofted dust from the red earth region shortly after onset. Similar products are available as above, but with better spatial resolution and slightly difference spectral specifications (Figs 9-13). Recall, the I bands provide the highest spatial resolution (375 m), while M band imagery is 750 m.
Continuing the GOES-East IR-based detection animations into the overnight hours (SWD in Fig 14, Dust RGB in Fig 15), detection becomes difficult as the lofted dust becomes increasingly dispersed, and cloud cover increases. However, careful analysis of the imagery allows one to diagnose the plumes extending northeast across southern Texas through the evening, with the faint remnants of the southern plume making it well up the Texas Gulf Coast.
Feature relative animations (such as when using feature following zoom in AWIPS) provide an intriguing alternative for viewing features in satellite imagery. Such features may include thunderstorms, boundaries, snow bands, dust, and smoke. Closed cellular convection is yet another feature for which a feature relative animation allows for a clearer picture of relevant processes. From the example on 4 Feb 2020 over the eastern Pacific, the divergence from the center of each cell is obvious, painting a picture of the implied rising air in the center of each cell (higher reflectance areas), and sinking air around the edges (low reflectance).
The same features are highlighted in Figure 2, but over a static region.
A zoomed out view of the region shows the development of the above closed cellular convection (in the middle of the scene) within a broader eastern Pacific Ocean anti-cyclone (Fig 3). Open cellualr covnection is also present in this scene, further to the southeast.
A longer animation (Sunday – Tuesday) shows the full evolution of the features behind the early week US trough and surge of cooler air and within the building eastern Pacific anticyclone (Fig 4).
A mid-upper level trough brought a variety of weather to the US during the first week of Feb 2020.
On 31 Jan, GOES-West WV imagery captured the early evolution of the trough over the central Pacific while a ridge was well established over the western US (Fig 1). Blending the satellite imagery with a model forecast not only allows one to guage model performance, but also provides a visualization of how features apparent in the imagery may evolve into the forecast period.
By 3 Feb, The ridge had eroded and the trough had advanced into the Great Basin, developing into a closed low (Fig 2). The periodic image degradation is due to the GOES-17 cooling system anomaly and and approaching vernal equinox.
The trough had sent a cold front south down the high plains during the overnight hours of the 2-3 Feb, with associated gravity wave perturbations evident in water vapor imagery (Fig 3). Combined with surface obs and RAP surface analyses, rapid pressure rises are apparent in the wake of the front, in addition to winds becoming northerly and then easterly and temperatures dropping considerable.
Nighttime Microphysics RGB imagery from GOES-West shows the rapid development of low clouds (dull yellow – green) across the high plains as temperatures dropped behind the front and the low levels became saturated. Surface obs also show winds becoming easterly upslope through the evening (Fig 4). Snowfall across the eastern plains to this point had been purely stratiform due to a saturated low layer and easterly upslope flow, per analysis of the RGB imagery.
After sunrise, cloud analysis is best done using the Day Cloud Phase Distinction RGB (Fig 5). The widespread low stratus deck (cyan) is obvious across the scene and contrasts with snow cover (green), high clouds (red), and clear ground (dark blue). By the end of this period, the upper low had advanced into northeast Utah, spreading stronger large scale forcing east across the Rockies, leading to increasing coverage of convective snow showers (textured reds).
A vigorous shortwave trough dug southeast out of Canada starting the evening of 19 Jan, continued across the middle of the US from the 20th to 21st, and across and east of Florida from the 21st to the 22nd. A long loop of GOES-East upper level water vapor imagery highlights the evolution and the shortwave as it dives southeast, with sinking and drying air (warm colors) on it’s western periphery, and rising and moistening air (white to green colors) to it’s east (Fig 1).
Zooming in to the southeast US as the storm moved offshore during the night of the 21st to the morning of the 22nd, significant strengthening of the shortwave is observed as thunderstorms in the ascending region and drying in the descending region both become more pronounced (Fig 2). Relevant large scale features are highlighted at the end of the period.
An alternative view of the strengthening is provided in the Airmass RGB imagery from GOES-16. The shades of red becoming more apparent on the western side of the storm represents sinking/drying/higher PV air into the upper troposphere (Fig 3).
The upper trough was accompanied by anomalously dry air and cold temperatures for the southeast US. Cooling temperatures can be visualized by the GOES-16 Land Surface Skin Temperature product, with skin temperatures across south Florida falling from the 70s during the day of the 21st to 30s during the evening (Fig 4).
The dry air is simialrly captured in the GOES-16 TPW product, with values across much of south Florida dropping to near or below 0.3″ (Fig 5).
This drying was also captured in radiosonde data, with the 12Z sounding from Key West measuring 0.3″ of TPW (Fig 6), which is well below average (1.2″) and actually is a new daily min for that location (Fig 7).
A surface low and intense and nearly stationary convection associated with the shortwave was captured in GOES-16 visible imagery and GLM Flash Extend Density data during the day on the 22nd (Fig 8).
A fast-moving shortwave trough brought a quick shot of wintry precip to parts of the Upper Midwest during the early part of 15 Jan. With an associated surface low pressure departing into the Ohio Valley, and high pressure building in from the west, northwest winds gusting to around 30 mph developed across the region.
The gusty northwest winds lofted snow off of the ground, causing areas of blowing snow and leading to reduced visiblities under otherwise clear skies across southern Minnesota and northern Iowa during the afternoon and early evening of 15 Jan. Bands reminiscent of cloud streets associated with the areas of blowing snow could be diagnosed in GOES-16 imagery. First analyzing the 0.64 um VIS, the bands of lofted snow area not obvious, but shadowing, especially later in the day, and the high (500 m) resolution of the band allowed for detection of the features (Fig 1).
The 1.6 um “snow/ice” band has been shown to effectively highlight areas of lofted snow (blowing snow) from the clear sky. The reflectance of the lofted tiny ice particles is typically between that of snow/ice on the surface (lower reflectance) and low liquid clouds (higher reflectance). Therefore, the lofted ice (over snow) can be more easily identified in the 1.6 m band vs the 0.64 um band, despite having a lower (1 km) spatial resolution.
Combining the aforementioned two bands with the 10.3 um IR window band in the Day Cloud Phase Distinction RGB, we get an even better picture of the blowing snow. The lofted ice is identified as a dull cyan above the surface snow (green) and adjacent to the low liquid water clouds (bright blue).
The Day-Snow-Fog RGB, which also includes the 1.6 um band (along with 0.87 um and 3.9 – 10.3 um difference), has been discussed as an effective means of identifying and tracking blowing snow. The contribution from 1.6 um band is the main driver behind the detection, with slight contribution from the other two components. For blowing snow detection, we replace the 0.87 um component with the higher resolution 0.64 um visible channel in order to capture finer details in the blowing snow bands.
Slight modifications were made to the RGBs in order to better highlight the feature of interest. The recipes used in this post are shown in figure 5.
Finally, a SNPP overpass during the early afternoon allowed for high resolution VIIRS imagery of the blowing snow bands to be collected. The 375 m 1.61 um band (I3) provided a great depiction of the blowing snow “streets” across the region.
A compact shortwave trough brought widespread severe weather to the southern US 10-11 January 2020. A Moderate Risk for severe storms was issued by the Storm Prediction Center for 10 January for a significant severe wind gust threat, with secondary threats of large hail and tornadoes. The threat on the 10th was centered over Texas, Oklahoma, Arkansas, and Louisiana, shifting east to Mississippi/Alabama on the 11th.
GOES-West upper-level water vapor imagery during the 48 hours leading up to convective development on the 10th show the evolution of the shortwave trough from off the northwest US coast, south along the west coast, and then east across the southwest US and into Texas (Fig 1).
The potent system drove enhanced low-level moisture into the region along with strong mid-upper level winds. This VISIT blog post highlights the utility of layer PW products for tracking the evolution of moisture and the EML leading up to the convective event.
Convection was ongoing/developing across Oklahoma during the morning/early afternoon on the 10th. Further south into Texas, convection developed a little later during the afternoon as the upper trough and associated strong forcing shifted closer, and inversion associated with the EML eroded.
By the late morning, GOES-16 derived motion winds indicated a swath of 60-70 knot SSW winds around 6 km AGL out ahead of the trough across north Texas (prior to CI here). Using the satellite-derived winds in combination with surface obs below to compute a vector difference, we achieve 0-6 km bulk shear values of around 55 knots, supporting the development of supercell thunderstorms (Fig 2). Further, the strong unidirectional deep layer flow is parallel to the surface/low-level boundary as diagnosed in visible imagery and surface obs (SSW to NNE). This orientation along with the strong shortwave forcing imply a likelihood of linear thunderstorm mode and wind threat quickly upon development.
GOES-16 1-min Day Cloud Phase Distinction RGB imagery showed widespread low stratus clouds (cyan) and showers (darker green indicating ice in cloud) across north Texas by early afternoon (Fig 3). During this period, convection rapidly develops southward across the region, clear in the 1-min RGB imagery per the vertical growth and transition to reds and yellows (result of cooling in the IR and increasing reflectance in the VIS as convection deepens).
Now looking at 1-min VIS/IR sandwich imagery for the following hour, convection continues to grow quickly southward across north Texas (Fig 4). The sandwich imagery maintains the high resolution/detail available in the 500 m VIS, while also representing the degree of cooling via the IR. Important storm to features are also easily detectable, including overshooting tops and above anvil cirrus plumes.
During this same period, VIS/GLM FED sandwich imagery showed lightning jumps with the strongest storms embedded in the broader line during early development (Fig 5). The north Texas storm for which a severe thunderstorm and tornado warning was issued saw lightning activity increasing from around 20 fl/5-min to around 90 fl/5-min within a ~15-minute period leading up to warning issuance. This rapid increase in flash density signifies a rapidly intensifying updraft. This particular storm had 1″ hail reported with it.
Analyzing 1-min visible imagery even further south, storms continued to develop south along a narrow line. The low sun angle allows for easy diagnosis of vertical growth as well as relevant storm top features such as overshooting tops and eventual above anvil cirrus plumes. Ahead of the developing storms, billow cloud formations indicate the continued presence of a stable layer inhibiting surface base convection.
A series of strong upper jets and related large scale subsidence helped to bring gusty winds to northeast Colorado for an extended period from 5-7 January 2020. Winds across the eastern plains generally gusted as high as 30-50 mph, while winds in the foothills and mountains peaked above 60 mph, with some reports of winds in excess of 80 mph.
During the afternoon of the 6th, gusty winds across the plains in the presence of dry, dormant fields resulted in widespread blowing dust. Recall the 10.3 – 12.3 um “split window difference” is able to highlight lofted dust quite well given the absorption of radiation at 10.3 um by silicate particles present in the dust. When lofted dust is present in the low levels, therefore, the 10.3 um band will sense slightly higher in the atmosphere, or cooler temperatures, compared to the 12.3 um band, resulting in near 0 to slightly negative SWD. In clear sky regions, the SWD will typically be positive since the 12.3 um channel will sense higher in the atmosphere (cooler temperatures) compared to at 10.3 um given the weak sensitivity of the 12.3 um “dirty window” channel to atmospheric moisture.
Similarly, one could view a 10.3 minus 11.2 um channel difference to identify lofted dust. While the difference may be greater in areas of dust, the clear sky difference will be less, reducing contrast between lofted dust and clear sky regions.
For this event, the lofted dust is readily apparent in the SWD across NE Colorado as darker shades of gray (Fig 1). I prefer a linear gray scale color table when using the SWD, whether it be for detecting lofted dust or moisture gradients. For this case, the range of the gray scale was adjusted to -2 to 5 to better highlight the lofted dust. It is recommended and simple to adjust the range on the fly to make the feature of interest stand out. However, a default value from around -3 to 10 will at least prove useful for initial detection.
The SWD is combined with the 10.3 um IR-Window channel and 11.2 µm – 8.4 µm Split Cloud Phase Difference (SCPD) to give us the Dust RGB, which can also be used for cloud classification. Similar to the SWD, the SCPD will highlight low-level lofted dust (small positive values). For this case, the dust is highlighted in the default RGB, as pink (Fig 2).
A slight modification to the RGB increases contrast and allows for easier identification of the dust plume. Instead of pink, the dust now appears as a dark blue above a brighter cyan/blue background (Fig 3). Low clouds appear as bright green, very cold surface as slightly darker green, and warmer surface as brighter cyan. High clouds are red.
The difference in RGB recipe is shown in Figure 4.
There is a GOES-R derived product for dust detection. In this case, it appears to capture the most opaque region of the dust plume (Fig 5).
That night, strong winds continued across the front range, extending into the adjacent plains. Surface observations are unable to capture the spatial intricacies of the gap winds as they seep into the lower elevations and I-25 corridor. Overnight GOES IR imagery showed the extent of the gusty winds via the presence of the warm anomalies due to the strong sinking/warming flow from west to east off of the high terrain (Fig 6).
In this example the warmer brightness temperatures are represented by darker shades of gray, while the cooler temperatures are lighter shades of gray. Gusty westerly winds per the surface obs match up with the warmer brightness temperatures per the satellite IR. The eastern slopes/downsloping region of the foothills appears warmer, while areas further west in the mountains are colder. Near the I-25 corridor, including the Denver area, localized areas of warming/gusty winds are diagnosed extending from west to east out of the mountain gaps. For example, early in the loop, Denver obs within a warmer plume indicate temperatures in the low 40s with winds gusting near 40 mph. Nearby areas with cooler IR brightness temperatures have surface obs in the low 20s and light winds. A broader area of gusty westerly downsloping winds and resulting warmer temperatures is measured over southern Wyoming advancing southeast into far northern Colorado, while much colder temperatures and lighter winds under the inversion are present over much of the rest of eastern Colorado.
An IR animation with RAP model temperature analysis highlights the broader surface temperature pattern, but fails to capture the smaller scale warmer gap flow regions (Fig 7).
The hourly Land Surface Temperature (LST) derived product form GOES-16 captures the smaller-scale warm features (lighter blue) quite well as they emanate from the foothills (Fig 8).
This more detailed (spatially/temporally) information available from GOES could aid mountain area forecasters in making their short-term forecast updates for wind and temperature during overnight wind events.
A snow squall developed off of Lake Ontario during the morning of 18 Dec 2019 ahead of a southward surging surface cold front and related upper trough digging into the northeast US (Fig 1). As has been noted, these localized bands of heavier snowfall can be diagnosed in the Day Cloud Phase Distinction (DCPD) RGB from ABI. This can be useful to forecasters as a supplement to radar, particularly for developing snow bands and over lakes, as well as when radar data are unavailable, for use in conjunction with surface observations and webcams.
In this case, the DCPD RGB highlighted the development of the band over Lake Ontario, and it’s evolution southeast into New York (Fig 2). The 1-min imagery allowed for real-time analysis of the band as it evolved. The bright green areas indicate the likely areas of heavier snow, resulting from the relatively high reflectance in the 0.64 um band and cooler brightness temperatures in 10.3 um band (taller/convective clouds), but low reflectance in 1.61 um band (ice at cloud top). Convective cloud elements are also detectable along the snow band in the DCPD RGB owing to the high (500 m) resolution from the 0.64 um band component. The initial snow squall warnings were issued at the end of this animation. The Rochester ob indicates heavy snowfall and significantly reduced visibility at the end of the loop as the snow band moved overhead. GOES-16 DMWs in the SFC-900 mb layer over the lake indicated low-level cloud motion of 25-30 knots, comparing well to winds observed at the surface. A meso-low is also apparent over the far eastern portion of the lake.
It should be noted that the default settings for this RGB were modified to better highlight the features of interest for this case given the relatively cool airmass and low light. The ranges used were (Red: -63.5, -2.5…Green: 60, 0…Blue: 45, 1), while the gamma’s all remained at 1.
For the final time-step in the above animation, a qualitative comparison between the DCPD RGB and MRMS composite refelctivity reveals a correlation between the bright green areas and highest reflectivity (Fig 3) across New York and Pennsylvania. Additional snow squall warnings were issued for the bands in northeast Pennsylvania.
The DCPD RGB and radar imagery are compared again in a two panel for the hour immediately following the animation in Figure 2 (Fig 4).
Figure 5 includes a long radar loop with the aforementioned NWS Snow Squall Warning polygon.
Ground truth from a Rochester webcam revealed rapidly deteriorating conditions associated with the snow squall (Fig 6).