Supercells were expected to initiate along a dryline in west-central Oklahoma during Saturday evening on April 23rd. The Storm Prediction Center had issued a slight risk of severe storms in their 1630 Z (1:30 PM CDT) outlook, with the risk tornadoes (5%), damaging hail (15%), and damaging wind (15%). Thunderstorms initiated around 2200 Z (5:00 PM CDT) as shown from the Day Cloud Phase Distinction RGB animation below. As the sun began to set near the end of the animation, decreasing contributions from the green (Channel 2, visible) and blue (Channel 5, near-IR) bands created a shift to more red colors in the imagery (Channel 13, clean-IR).
Animation of the Day Cloud Phase Distinction RGB from 2130 to 2300 Z (4:30-6:00 PM CDT).
An SPC Mesoscale Discussion and Tornado Watch, along with NWS Norman Public Information Graphics show the transition from the initial SPC Convective Outlook to the warnings that would later be issued that evening. (Images below in chronological order)
As convection matured into supercells after sundown, satellite imagery became confined to the infrared bands (Channels 7-16), with Clean-IR imagery most often used. Additionally, rapidly updating (1 minute) lightning data from the GLM Flash Extent Density product can provide information about thunderstorm trends between NEXRAD full-volume scans (4-5 minute updates). At night, the GOES-16 GLM detection efficiency often exceeds 90% across the south-central United States.
Intensification of two supercells and tightening of their low level mesocyclones, southwest of Oklahoma City and southwest of Stillwater, as indicated by radar prompted the NWS Norman office to issue tornado warnings for both storms. The Tornado Warning for the Stillwater supercell was issued at 2359 Z (6:59 PM CDT), and the Tornado Warning for the Oklahoma City supercell was issued at 0003 Z (7:03 PM CDT).
5-minute ABI and GLM data from 2330 to 0030 Z (6:30 – 7:30 PM CDT) Left: GLM Flash Extent Density (5 minute total) overlaid on the ABI Clean-IR band. Right: ABI Clean-IR Band.
The animation above is from 2330 Z to 0030 Z (5 minute intervals), and shows how both storms intensified from the perspective of the GLM FED and ABI Clean-IR products. Deep overshooting tops were observed from the ABI along with notable increases in GLM flash rates. In this scenario satellite information may have provided a ‘heads-up’ on which storms to monitor, along with additional confirmation of trends observed from NEXRAD.
One-minute data was observed from the GOES-East Mesoscale Domain for both products (below). In this scenario NWS Norman also had access to the Terminal Doppler Weather Radar at the Oklahoma City Airport (TOKC), providing 1-minute radar reflectivity and doppler velocity data within the vicinity of the airport. For the supercell near Oklahoma City, this may make a forecaster less reliant on one-minute satellite data when making warning decisions. However, for the storm southwest of Stillwater no TDWR data was available. The rapid increase in lightning flash rates identified by the GLM FED product for this storm can provide additional verification for an NWS forecaster that the updraft was intensifying, and tightening of the low level mesocyclone prior to tornadogenesis may be imminent.
1-minute data from the GOES-East ABI (Clean-IR) and GLM (FED) within the mesoscale domain. Imagery is from 2350 – 0010 Z (6:50-7:10 PM CDT) (Click animation to view at full size)
Thunderstorms and their associated severe winds have a long history of impacting aircraft. Many tragic incidents demonstrate how avoiding the turbulent updrafts and downdrafts within thunderstorms is paramount to protect the aircraft and ensure the safety of the passengers and crew. Convective activity can significantly alter flight paths for both commercial and general aviation. Airlines incur costs in the tens of millions of dollars each year in extra fuel, diversions, and labor costs to manage thunderstorm-related impacts and delays. This case study uses GOES-R series satellite imagery to examine severe weather and its relationship with the resultant flight paths of nearby aircraft.
In mid-May 2019, a strong upper-level trough was present over the western United States, advecting positive potential vorticity over the Central High Plains and the upper Great Plains (Fig. 1). Upper-level support for ascent was combined with low-level support in the form of a developing leeside surface cyclone advancing northeastward across northeast Colorado and into northwest Kansas and southwest Nebraska. A mesoscale dry line front developed along the downslope terrain of eastern Colorado and western Kansas, providing strong localized support for convective initiation coupled with thermodynamic atmospheric parameters supportive of further convective development. This eventually resulted in thunderstorms and severe weather in the form of tornadoes, large hail, and damaging winds.
Fig. 1: CONUS imagery from GOES-16 of the Channel-9 Water Vapor imagery from the Central United States. A dry line front is clearly visible as drier air (yellow shades) advancing eastward through eastern Colorado.
The National Centers for Environmental Prediction (NCEP) Central Operations stewards a meteorological observational archived database called the Meteorological Assimilation Data Ingest System, or MADIS. MADIS incorporates data from a wide variety of sources, including international, federal, state, and local agencies, universities, volunteer networks, and the private sector. Part of this private sector data included within MADIS is the location, time, heading, and altitude for registered aircraft within US airspace reporting via the Aircraft Communications Addressing and Reporting System (ACARS). Aircraft data from MADIS is encrypted such that flights specifics (i.e., tail number, aircraft operator, etc.) are not seen by the user, but the archived locations and headings of registered aircraft are available. While this dataset takes considerable care to develop into a GIS-friendly product, it can be a powerful tool to demonstrate the impacts weather has on aviation.
For this severe weather event, convective initiation was a core component of the forecast. Thunderstorms were expected to mature quickly as insipient updrafts were able to break through a relatively strong capping temperature inversion. The Geostationary Lightning Mapper (GLM) supplies valuable satellite confirmation of convective initiation by quickly highlighting the initial flashes and their spatial extent with gridded imagery. The GLM lightning observations pair well with GOES-16 ABI imagery, including the Day Cloud Phase Distinction RGB, which can highlight the development of ice and charge separation within clouds. Please note, the Day Microphysics RGB, which has notably different ingredients and recipe from the more familiar NWS AWIPS Day Cloud Phase Distinction RGB, but shares a similar purpose, is shown in this blog post. In this RGB and relevant to our discussion in this case, convective initiation can be diagnosed as cumulus clouds grow and transition from shades of tan/pale yellow to deeper red/orange.
From NWS Goodland at 1919 UTC: “GOES-16 Day-Cloud Phase Distinction imagery indicating a few failed attempts at thunderstorm initiation along dry line and with mesoscale data supporting ML inhibition between 50 in the south to 150 j/kg in the north. Will likely take another hour of insolation to further weaken capping, which will coincide with approaching pv anomaly from the west. Expect strong storms to fire between 2 and 3 pm MDT. Relatively high LCL’s and weak 0-1km shear support initial threats being very large hail greater than 2 inch and diameter. Tornado threat is somewhat low, but should increase as storms move east and north towards triple point near NE border and as LLJ increases around 6 PM.”
Overlaying the ACARS flight data with the GOES-16 Day Microphysics RGB during the early afternoon hours before initiation along the dry line near the Colorado/Kansas border shows a busy corridor of aviation traffic arriving at and departing from Denver International Airport (DEN) as well as many other busy regional airports in the area (Fig. 2). It is also apparent how planes are instructed to fly along arranged routes from point to point, and how those routes can be adjusted on the fly by weather. A considerable amount of traffic traveling across the domain appears to have taken off from and is destined for airports which are not displayed here. [The destinations and origins of the aircraft are not specified in this dataset available to the public.]
Fig. 2: Mesoscale Sector and CONUS imagery from GOES-16 of the Day Microphysics RGB from the Central High Plains in the time leading up to convective initiation. Overlaid is the Flash Extent Density gridded imagery from the GLM and the reported locations and headings from archived ACARS flights.
The failed attempts at convective initiation in the early afternoon as noted by NWS Goodland can be seen on the RGB imagery, and it finally gave way to thunderstorm activity around 2020 UTC. Just before the first lightning strikes are observed by the GLM, a distinct, robust turret confirming the presence of a building updraft can be seen in the RGB imagery just southwest of the GLD airport (Fig. 2). At the same time, more widespread red pixels, associated with an increasing amount of ice present at the top of the cloud, are shown in the RGB imagery just before lightning was first observed with this particular storm by the GLM.
After convection began in earnest along the dry line, it is clear that the rapidly developing weather conditions were having a significant impact on aviation by affecting the flight paths of hundreds of aircraft in this area. However, when looking at Channel-13 Longwave Infrared ABI imagery alone (Fig. 3), it is difficult to ascertain distinct trends in the flight paths of aircraft when compared to brightness temperatures at the top of the cloud. The gridded products from the GLM provide insights into the spatial extent of lightning with satellite-based lightning observations with 1-minute updates over much of the full disk of both GOES-East and GOES-West. The spatial extent of lightning can be investigated with the Flash Extent Density GLM gridded imagery through supercell development as in Figure 4.
Fig. 3: Mesoscale Sector and CONUS imagery of the Channel-13 Longwave IR ABI from GOES-16. Overlaid is the reported location and headings from archived ACARS flights.
The GLM provides unique observations of the spatial extent of lightning, and in this context demonstrates how aircraft are diverting around ongoing lightning flashes within thunderstorms as they are occurring. The convective cores of the thunderstorms can generally be identified by the more numerous flashes occurring within or near the updraft. Greater flash counts manifest as warmer colors in Flash Extent Density imagery, and vice versa. Due to the relatively low flash rates in this event, the standard color scale for Flash Extent Density may not always highlight convective cores.
Fig. 4: Mesoscale Sector and CONUS imagery from GOES-16 of the Day Microphysics RGB from the Central High Plains from before convective initiation through 00Z on 18 May 2019. Overlaid is the Flash Extent Density gridded imagery from the GLM and the reported locations and headings from archived ACARS flights.
Another GLM product which may provide additional value in identifying the location of rapidly evolving convective updrafts is Minimum Flash Area, which reports the area in square kilometers of the smallest lightning flash observed by the GLM within each pixel in the specified time window. In addition to being more numerous, flashes which are more associated with strengthening or mature updrafts are typically smaller in spatial extent. This occurs because of the large amount of charge separation taking place over a spatially small area within the vertically oriented updraft. Thus, Minimum Flash Area highlights these updrafts, with warmer colors indicating smaller minimum flash sizes, and vice versa. Minimum Flash Area alone can provide a significant amount of context for immediate impacts on not only aviation but also severe weather warnings and reports (Fig. 5).
Fig. 5: GOES-16 GLM Minimum Flash Area gridded imagery overlaid with NWS Severe Thunderstorm (yellow) and Tornado (red) warning polygons, CWSU SIGMET advisories (orange polygons), severe weather reports from the SPC archive (with similar symbol styling), and the reported locations and headings from archived ACARS flights. Note that there may be some spatial offset between the satellite-based GLM observations and ground-based observations approaching the edge of the field of view of the GLM; read more here.
Aircraft take a circuitous route on this day to avoid thunderstorm activity, if possible. Flash Extent Density alone may not necessarily give the full perspective on which lightning strikes pilots and air traffic control may be actively avoiding. Minimum Flash Area imagery provides additional context on which flashes are more likely associated with core convective updrafts and which are more likely associated with the anvil/stratiform regions by classifying flashes by the size of their spatial extent. In this imagery and in other cases, airplanes tend to more acutely avoid the more numerous, smaller flashes within the turbulent updrafts, and less so avoid the less frequent, larger, anvil/stratiform region flashes.
Incorporating the GLM gridded products of lightning observations into aviation forecasting operations can be a vital tool for meteorologists when spotting convective initiation and turbulent updrafts by classifying lightning flashes by their spatial extent. This additional information about convective activity can more precisely guide aviators through natural hazards such as thunderstorms, snow squalls, and other atmospheric phenomena.
Satellite Liaison Blog 