ProbSevere products over the Southern Plains

May 3rd, 2021 |

The NOAA/CIMSS ProbSevere portfolio contains AI models for nowcasting convective weather. I’ll use Monday’s severe weather over the Southern Plains to highlight several of them.

A strong cold front spawned numerous severe-hail, wind, and tornado producing storms over Texas and Oklahoma, aided by very large values of convective available potential energy (CAPE; > 4000 J/kg).  You can see numerous storm reports in Figure 1.

210503_rpts Reports Graphic

Storm Prediction Center’s preliminary severe storm reports for May 3rd, 2021.

Probsevere version 2 (PSv2) is an operational set of models at NOAA, which predict the probability of severe hail, severe wind, and tornadoes, in the next 60 minutes. The models are storm-centric, and the models’ domain is the entire contiguous United States (CONUS).  These models use MRMS (radar), GOES (satellite), short-term NWP, and terrestrial-based lightning observations to generate probabilistic guidance of severe hazards. Figure 2 shows output from an experimental version (PSv3), which includes additional MRMS, GOES, and NWP fields as predictors in a machine learning model.

Figure 2: ProbSevere v3 contours (colored, around storms), MRMS MergedReflectivity, and NWS severe weather warnings (yellow and red boxes) for storms over the Southern Plains. The second outer contour around some storms is colored by the probability of tornado.

 

Another ProbSevere product is a convolutional neural network that uses GOES-R ABI and GLM images to detect regions of intense convection, and is often correlated with strong overshooting tops, “bubbly-like” texture in visible imagery, strong lightning cores, and the cold-U/above-anvil cirrus plume signature. The intense convection probability (ICP) can be run on the 1-minute mesoscale scans as well as 5-minute CONUS sector scans aboard the GOES satellites. The ICP does not require radar data, and may also be able to operate on data from satellites with similar intruments (e.g., Meteosat Third Generation). ICP output is being used as a predictor in the experimental ProbSevere v3.

 

Predicting when and where lightning will occur is also important for many users, such as mariners, aviators, and outdoor event managers. The probability of lightning model (PLTG) is also a convolutional neural network, using images of visible, near-infrared, and longwave-infrared channels to nowcast lightning occurrence in the next 60 minutes. The purple-to-orange shaded regions in the video below show GLM flash-extent density (i.e., flashes passing through a location).

NOAA/CIMSS ProbSevere with a tornado in Tallahassee, FL

January 27th, 2021 |

NOAA/CIMSS ProbSevere display, 1545 – 1700 UTC on 27 January 2021 (Click to animate)

A tornado struck the Tallahassee, FL, airport at 1643 UTC on 27 January 2021 (SPC Storm Report).  The animation above shows ProbSevere (version 2) fields (from this site) in the hour leading up to tornadogenesis.  The animation demonstrates how ProbTor values can be used to identify for closer scrutiny a particular radar object:  the radar object that ultimately caused a tornado showed greater ProbTor values (than surrounding identified radar objects) in the hour leading up to tornadogenesis. In addition, ProbTor values ramped up quickly just prior to tornadogenesis as low-level azimuthal shear jumped.

One time series below compares ProbWind, ProbHail and ProbTor for the radar object (#15080) that produced the tornado; for this event, ProbWind and ProbTor values were comparable until a ramp-up in ProbTor values before the tornado occurred. The second time series shows the various components of ProbTor for radar object 15080 (both time series courtesy John Cintineo, SSEC/CIMSS).  Note in particular that this storm was not a lightning-producer.  Much of ProbTor’s variability was determined by changes in low-level azimuthal shear.

NOAA/CIMSS ProbSevere values (ProbWind, ProbHail, ProbTor) for radar object #15080, 1530 – 1658 UTC on 27 January 2021 (Click to enlarge)

NOAA/CIMSS ProbTor and component values for Radar object #15080, 1530 – 1658 UTC on 27 January 2021, associated with the Tallahassee FL tornado (Click to enlarge)

Lead time with ProbTor in this example was not exceptional.  However, its elevated values in the hour leading up to the tornado could have provided better situational awareness, and perhaps enhanced confidence in warning issuance for this well-warned event.

_________________________________________________________________________________________________________

GOES-16

GOES-16 “Red” Visible (0.64 µm, left) and “Clean” Infrared Window (10.35 µm, right) images, with plots of SPC Storm Reports [click to play animation | MP4]

Unfortunately, the default Mesoscale Domain Sectors were positioned too far north to cover the Florida Panhandle — but 5-minute CONUS Sector GOES-16 (GOES-East) “Red” Visible (0.64 µm) and “Clean” Infrared Window (10.35 µm) images (above) depicted a west-to-east oriented line of thunderstorms across the northern portion of the Panhandle; a trend of cooling cloud-top infrared brightness temperatures was seen as the convection began to produce the tornado.

There was an overpass of the Terra satellite about 19 minutes before the start of the tornado event, at 1618 UTC — 1-km resolution MODIS Visible (0.64 µm) and Infrared Window (11.0 µm) images are shown below.

Terra MODIS Visible (0.64 µm) and Infrared Window (11.0 µm) images [click to enlarge]

Terra MODIS Visible (0.64 µm) and Infrared Window (11.0 µm) images [click to enlarge]

NOAA/CIMSS ProbSevere with a tornadic cell in Kansas/Oklahoma

April 26th, 2016 |
GOES-14 Visible (0.63 µm) Imagery, 26 April 2016. An orphan anvil is indicated by the Green Arrow at the start of the animation (click to play animation)

GOES-14 Visible (0.63 µm) Imagery, 26 April 2016. An orphan anvil is indicated by the Green Arrow at the start of the animation (click to play animation)

April 26 2016 was a day of well-anticipated severe weather (even a week out!) over the central and southern Plains, with a Moderate Risk of Severe Weather predicted for parts of Nebraska, Kansas, Oklahoma and Texas. The GOES-14 visible animation, above, shows the development of strong thunderstorms in north-central Oklahoma that propagated into south central Kansas, producing hail around 2000 UTC. Note the presence of an orphan anvil just downstream of the developing convection (to the south of the Green Arrow) at the beginning of the GOES-14 SRSO-R animation (that unfortunately has a 15-minute data gap starting at 1900 UTC).

How did the NOAA/CIMSS ProbSevere product perform with this severe cell? ProbSevere provides a probabilistic estimate of whether a cell will produce severe weather within the next 60 minutes. The animation below shows the quick development of the radar feature that became the hail producer. The Satellite Growth of this particular storm was not observed to be strong. Moderate satellite growth and weak glaciation was diagnosed. However, ProbSevere values became very large because of the environment in which the cell developed, because of the presence of large MRMS MESH observations, and active lightning. ProbSevere exceeded a 50% threshold at 1912 UTC, 6 minutes before the Severe Thunderstorm Warning was issued.  The Table at the bottom shows the ProbSevere components as a function of time.

According to SPC storm reports, the cell produced a brief rope tornado at 2058 UTC in far southern Kansas. This storm was blogged about at the Hazardous Weather Testbed. Click here and here for blog posts on the environmental instability.

NOAA/CIMSS ProbSevere Output, 1824-1946 UTC on 26 April 2016 (click to play animation)

NOAA/CIMSS ProbSevere Output, 1824-1946 UTC on 26 April 2016 (click to play animation)

A zoomed-in animation of the Visible Imagery shows the orphan anvil developing around 1740 UTC. (A rocking animation is here).

GOES-14 Visible (0.63 µm) Imagery, 26 April 2016. The orphan anvil is indicated by the Cyan Arrows through the animation (click to play animation)

GOES-14 Visible (0.63 µm) Imagery, 26 April 2016. The orphan anvil is indicated by the Cyan Arrows through the animation (click to play animation)

 

Time (UTC) ProbSevere MUCAPE Env. Bulk Shear MRMS MESH (Inches) Satellite Growth Satellite Glaciation # Flashes
1854
1858 20% 4739 41.9 0.29 1.9% (Moderate) 0.02 (Weak) 0
1900 29% 4702 41.8 0.45 1.9% (Moderate) 0.02 (Weak) 0
1908 34% 4640 40.9 0.54 1.9% (Moderate) 0.02 (Weak) 5
1910 47% 4628 40.7 0.65 1.9% (Moderate) 0.02 (Weak) 13
1912 59% 4623 40.4 0.65 1.9% (Moderate) 0.02 (Weak) 24
1914 58% 4619 40.1 0.65 1.9% (Moderate) 0.02 (Weak) 24
1916 58% 4614 39.8 0.65 1.9% (Moderate) 0.02 (Weak) 24
1918 54% 4614 39.8 0.60 1.9% (Moderate) 0.02 (Weak) 24
1920 60% 4592 39.4 0.74 1.9% (Moderate) 0.02 (Weak) 20
1922 65% 4591 39.1 0.80 1.9% (Moderate) 0.02 (Weak) 20
1924 73% 4591 39.1 0.80 1.9% (Moderate) 0.02 (Weak) 25
1926 75% 4572 38.8 0.84 1.9% (Moderate) 0.02 (Weak) 26
1928 88% 4578 38.7 1.01 1.9% (Moderate) 0.02 (Weak) 31
1930 89% 4578 38.7 1.01 1.9% (Moderate) 0.02 (Weak) 36
1932 97% 4580 38.6 1.24 1.9% (Moderate) 0.02 (Weak) 49
1934 97% 4560 38.3 1.24 1.9% (Moderate) 0.02 (Weak) 58
1936 97% 4544 38.1 1.24 1.9% (Moderate) 0.02 (Weak) 58
1938 97% 4543 38.0 1.24 1.9% (Moderate) 0.02 (Weak) 58
1940 97% 4540 37.8 1.26 1.9% (Moderate) 0.02 (Weak) 58
1942 98% 4528 37.7 1.53 1.9% (Moderate) 0.02 (Weak) 56
1944 99% 4516 37.5 1.71 1.9% (Moderate) 0.02 (Weak) 56
1946 99% 4507 37.4 1.71 1.9% (Moderate) 0.02 (Weak) 56

 

NOAA/CIMSS ProbSevere with a Nebraska Hailstorm

September 22nd, 2015 |
GOES-13 Visible (0.63 µm) images [click to play rocking animation]

GOES-13 Visible (0.63 µm) images [click to play rocking animation]

A severe hail-producing thunderstorm moved over northeast Nebraska before noon on 22 September (SPC Storm Reports). The region hit was just south of a Marginal Risk of Severe Weather (The update at 1630 UTC included the region of severe weather). The GOES-13 visible animation, above, shows the initial development occurring along a subtle cloud line aligned mostly east-west.

The NOAA/CIMSS ProbSevere model produces a probability that a developing thunderstorm will initially produce severe weather within the next sixty minutes. It consistently supplies information with a good lead time, and the storm on 22 September was no exception. The animation below shows the product for about an hour before the first storm report at 1408 UTC. The storm out of which the hail dropped was, at 1300 UTC, flagged as having a ProbSevere under 10%; values exceeded 10% at 1314 UTC and then jumped to 60+% at 1336 UTC (the first time that the value exceeded 50%) Values fluctuated between 60 and 80% between 1336 and 1400 UTC. After 1400 UTC, values increased into the mid-80s. The first report of hail was at 1408 UTC, 32 minutes after ProbSevere jumped above 50%. A severe thunderstorm warning for hail was issued at 1412 UTC.

NOAA/CIMSS ProbSevere values, 1300-1412 UTC on 22 September 2015 [click to play animation]

NOAA/CIMSS ProbSevere values, 1300-1412 UTC on 22 September 2015 [click to play animation]

The GOES Sounder Lifted index product, below, (also available here) showed the instability that was present over the central Plains.

GOES-13 Sounder DPI Values of Lifted Index [click to play animation]

GOES-13 Sounder DPI Values of Lifted Index [click to play animation]