ProbSevere and LightningCast over La Crosse, Wisconsin

June 15th, 2022 |

ProbSevere is a weather product that uses GOES-16 data to help predict the probability of any severe weather (hail, wind, or tornados) happening in the next 60 minutes. ProbSevere LightningCast predicts the probability that a GLM Lightning observation will occur in the next 60 minutes. Below are examples of these products over the Western Wisconsin area as a convective system rolls through La Crosse, Wisconsin.

The ProbSevere probability product imaged as contours, overlayed with GOES-16 Day Cloud Convection RGB for 06-15-2022 from 20:00 to 20:22Z near La Crosse, Wisconsin [Click to enlarge].
The LightningCast probability product imaged as contours of 75, 50, 25, and 10 percent, overlayed with GOES-16 Day Cloud Convection RGB for 06-15-2022 from 20:00 to 20:22Z near La Crosse, Wisconsin [Click to enlarge].

ProbSevere (and NUCAPS) with the Gaylord Tornado

May 20th, 2022 |
ProbSevere Version 3 readout at 5-minute intervals, 1800 – 2030 UTC on 20 May 2022 (Click to enlarge)

ProbSevere (version 3) is an online tool that is available at this link. It gives the probability that the designated radar object will product severe weather in the next 60 minutes, and it was designed to be used in conjunction with other radar, satellite and model data to increase confidence in warning issuance. It is available online here; ProbSevere (version 2) is also online — and available within AWIPS. The RealEarth-based readout, above, shows the radar object that produced the EF-3 Tornado in Gaylord. (Click here for the NWS Gaylord writeup). ProbSevere values were consistently high with the radar object thoughout the trek across the northern Lower Peninsula of Michigan. Here is an image showing the accumulated ProbSevere values across northern Michigan (from this RealEarth link)

Read-outs of PSv3 and PSv2 values are available at this link, and the plot for the radar object in the animation above (#170831) is shown below (Here is a permanent link that allows you to view values!). Of note here is that PSv2 values are larger than PSv3 values. This is because PSv3 is better calibrated. For example, the Gaylord Tornado has a ProbTor value (plotted in red) of 90%! However, it is not the case that ProbTor (version 2) predictions with values of 90% lead to tornadoes 90% of the time: ProbTor v2 is overpredicting values in that case. ProbTor (version 3) is better calibrated, and the lower values are more in line with observations. Users familiar with ProbTor version 2 will need to adjust their expectations when transitioning to ProbTor version 3 (or indeed from any of the ProbSevere version 2 products to ProbSevere version 3 products).

ProbSevere readout for values associated with Object ID# 170381 (click to enlarge)

NOAA-20 overflew the region of severe weather just before 1800 UTC (link). Gridded NUCAPS fields from this site, below, show marked instability over the northwestern Lower Peninsula of Michigan, especially in the 700-500 mb layer, where gridded values exceed 7 C/km. The gridded values also show a decrease in stability for any storms moving into the Lower Peninsula from the west; however, Quality Control flags (shown below in a toggle with 850-mb Temperature), show profiles that did not converge over Lake Michigan.

Gridded NUCAPS Lapse Rates, 850-700 mb. 700-500 mb and 850-500 mb (click to enlarge)
NUCAPS Quality Control Flags (Green: Retrievals converged; Red: Retrievals did not converge; Yellow: Microwave retrieval converged, infrared did not converge) and 850-mb Temperature at 1745 UTC on 20 May 2022 (Click to enlarge)

LightningCast in Tampa, Florida

April 8th, 2022 |

Thunderstorms were slowly but surely edging their way dangerously close to Raymond James Stadium in Tampa, FL, on April 2nd. The New York Yankees and Atlanta Braves had just completed a spring training game at the stadium, when two people were struck by lightning in one of the parking lots surrounding the stadium (they were hospitalized but reported to be in stable condition).

ProbSevere LightningCast is an experimental deep-learning model that is running in near-real time at CIMSS. It uses images of GOES-R Advanced Baseline Imager (ABI) visible, near-IR, and longwave-IR channels to predict the probability of lightning (as observed by the GOES-R Geostationary Lightning Mapper [GLM]) in the next 60 minutes.

Below is a time series of the LightningCast probability and GLM-observed lightning at and near Raymond James Stadium (left panel), along with an animation of LightningCast probability contours, GOES-16 0.64-µm reflectance (from a 1-minute mesoscale sector), and GLM flash-extent density (right panel) near the stadium (red circle). In this way, users can see how the model’s probabilities evolved over time at a specific location and within the vicinity.

Figure 1: Time series of LightingCast probability and GLM-observed lightning at Raymond James Stadium (left). Animation of LightningCast contours, GOES-16 visible reflectance, and GLM flash-extent density (right) near the stadium (red circle).

Police officers responded to the two individuals struck by lightning at 3:45 PM local time (19:45 UTC). Based on the footprint of the GLM flash-extent density, they were struck at approximately 3:32 PM. The LightningCast probability of lightning was 75% 30 minutes before the lightning strike (remaining mostly above 50% between 3:00 and 3:32 PM). The probability of lightning first reached 50% about 1 hour before the lightning strike, and lightning started occurring within the vicinity (within 25 km) about 45 minutes before the strike.

Output from LightningCast, which leverages the high spatial, temporal, and spectral information found in GOES-R ABI, can help objectively quantify the short-term threat of convective hazards such as lightning. The model could perhaps be used by forecasters to advise outdoor venues such as stadiums to take mitigating actions sooner, or by individuals to help make safe decisions.

Figure 2: Annotated time series of LightningCast probability of lightning and GLM observations near Raymond James Stadium.

Severe weather in South Carolina

April 5th, 2022 |
ProbSevere (version 2) over South Carolina, 1902-2012 UTC on 5 April 2022 (click to enlarge)

SPC Storm Reports from 5 April 2022 note a Tornado hit Allendale SC shortly before 2000 UTC. The animation above shows the ProbSevere (version 2 — available online here) display prior to and just after the tornado. NOAA/CIMSS ProbSevere identifies and tracks the radar feature associated with this tornadic storm. ProbSevere is designed to give forecasters more confidence in warning issuance. ProbSevere in this case highlights the radar object associated with model fields and satellite/radar observations that are most suggestive of a storm supporting tornadogenesis — note that ProbSevere values in adjacent cells are smaller. A meteorogram of this radar object, shown below (and available here), shows the ProbSevere components, and also ProbHail/ProbWind/ProbTor/Probsevere values, for both versions 2 and 3. ProbSevere values increased at around 1930 UTC. ProbSevere v3 values generally are smaller than ProbSevere v2 values; ProbSevere v3 will be demonstrated this year at the Hazardous Weather Testbed (HWT) this year. (Click here to view the Charleston, SC (KCLX) radar at 1958 UTC; a pronounced hook is apparent)

ProbSevere readout meteogram, Object Number 653760, 5 April 2022 (Click to enlarge). For the ProbSevere readout (top row), ProbTor values are in red, ProbHail values are in green, ProbWind values are in blue. (Click to enlarge)

This tornadic cell stood out in the visible imagery. GOES-16 Mesoscale Sector #2 on 5 April included portions of South Carolina. The mp4 animation below, from the CSPP Geosphere site, (this direct link to the animation will be valid for a bit less than a week) shows the tornadic cell erupting at around 1930 UTC near the Georgia/South Carolina border.

GOES-16 Visible Imagery (Band 2, at 0.64 µm) from Mesoscale Sector 2, 1900-2005 UTC 2022

Observations — satellite and radar — both showed the obvious storm. How did short-range guidance perform? A Polar Hyperspectral modeling system — also to be demonstrated at the Hazardous Weather Testbed — produces hourly 18-hour forecasts with initial fields influenced by Sounder Data from the Polar Orbiting satellites Suomi-NPP, NOAA-20, Metop-B and Metop-C. Infrared sounder (CrIS on Suomi/NPP and IASI on Metop) and Microwave sounder (ATMS on Suomi/NPP, AMSU/MHS on Metop) data can produce a more accurate initialization of the moisture distribution in atmosphere. The forecast initialized at 1200 UTC for Lifted Index and Significant Tornado Parameter, valid at 18, 19 and 20 UTC — that is, 6-h, 7-h and 8-h forecasts, below, shows increasing instability before the tornado in Allendale.

Lifted Index (left) and Significant Tornado Parameter (right) at 1800, 1900 and 2000 UTC on 5 April 2022 (Click to enlarge). Forecast initialized at 1200 UTC on 5 April 2022

The model runs initialized at 1700, 1800 and 1900 UTC, showing fields from initialization through 0000 UTC on 6 April 2022, are shown below. Note that the area with a Significant Tornado Parameter signal is mostly confined to southeastern South Carolina — that is, near the coast. Storm reports show that severe weather was mostly near the coast as well.

Lifted Index (left) and Significant Tornado Parameter (right) hourly from at 1700 UTC on 5 April 2022 through 0000 UTC on 6 April 2022 (Click to enlarge). Forecast initialized at 1700 UTC on 5 April 2022
Lifted Index (left) and Significant Tornado Parameter (right) hourly from at 1800 UTC on 5 April 2022 through 0000 UTC on 6 April 2022 (Click to enlarge). Forecast initialized at 1800 UTC on 5 April 2022
Lifted Index (left) and Significant Tornado Parameter (right) hourly from at 1900 UTC on 5 April 2022 through 0000 UTC on 6 April 2022 (Click to enlarge). Forecast initialized at 1900 UTC on 5 April 2022

These fields are also available in AWIPS via an LDM feed (in preparation for HWT). The 3 images below show changes in the 2000 UTC forecast (from the model initialized at 1400, 1600 and 1700 UTC). The trend towards higher Significant Tornado Parameter over southeastern South Carolina is obvious. Note that Allendale’s location is shown.

Significant Tornado Parameter valid at 2000 UTC on 5 April 2022 from forecasts initialized at 1400, 1600 and 1700 UTC (Click to enlarge)

Added: the GLM on GOES-16 saw a dramatic increase in Flash Extent Density with the tornadic storm starting around 1925 UTC on 5 April. (Click here to see a slower animation from 1920 – 1930 UTC)

GOES-16 GLM Flash Extent Density, 1900-2029 UTC on 5 April 2022 (Click to enlarge)