Week 1 Day 2 Operations

Today we will be operating across west and central Texas looking for developing storms along a slow moving front.

HWT Spring Experiment Begins

The 2019 combined satellite and radar Spring Experiment started today and will run for 6 of the next 7 weeks evaluating various satellite based products as well as some radar derived products. Today is generally a familiarization day as forecasters get set up to operate in forecast and warning mode over the next few days looking at some active weather hopefully. Look here for updates throughout the week on the satellite based products.

- Michael

NUCAPS in 4-Panel

When NUCAPS Sounding Availability is loaded in a 4-Panel and is 'Editable', you can sample the data by clicking on any of the other panels. This could be useful if there is an area of interest in one of your other datasets (i.e. vis/CAPE/PWAT) so that you don't have to find the nearest dot using the NUCAPS imagery; simply click the area of interest on the panel you're investigating. Works in panel combo rotate as well.

-- FLGatorDon

Precipitable Water Overview

Looking at the different precipitable water (PW) products available in the HWT and doing a quick overview the All-Sky products provides a great first guess to fill in the PW where it is cloudy. Both the Merged TPW and the All-Sky take the first step in filling in where there are clouds. The image below  shows the sheer volume of data that isn't available due to the pesky cloud cover. The 4-Panel to the left shows the All-Sky PW and CAPE on top vs. the raw derived PW and CAPE from GOES-16. On the right you can see the visible satellite and the All-Sky mask showing that most of the data, especially over Texas and Oklahoma is raw GFS (gray areas) at this point.



Looking at the Blended TWP vs. the All-Sky there are significant differences over north Texas and Oklahoma for this time frame. The BTWP product "is not forecast model dependent. ATPW uses  GFS model winds to advect the microwave retrievals and the GOES-16 component uses GFS in its TPW solution." You can see where the blended product (big window below) only shows about 0.75 in PW, while the All-Sky is showing 1.25 in. across the Norman WFO. This can make a big difference when looking at rainfall forecasting and trying to assess just how much moisture in the atmosphere is over an area. In this case would certainly lean towards the All-Sky and then compare the information to other model soundings (from the HRRR, NAM, ECMWF, etc.) and to actual Upper Air soundings to see how the areas populated by the raw GFS are doing.



-Alexander T.

ProbSevere: Color Table Modification

While I've been a big fan of the ProbSevere Model for some time now, the default color curve has always been a little challenging for me to differentiate between the different percentages. Trying to find the right balance between the radar color tables and ProbSevere I know can be tough, but here's my first go at attempting to better differentiate when the percentages move into the next 10% range. Unfortunately, I'm unable to really test this modified color table out since we're currently not getting radar data in from the Davenport area, so will reassess this once I'm able to overlay the model output with radar imagery. But, just having the colors pop a little more is already helpful to me! Oh, and I'm very appreciative that ProbSevere v.2 now includes the separated values (i.e. ProbWind, ProbHail, ProbTor). Looking forward to testing this out once there's a case to evaluate with it.

*Note: As of this posting, WFO DVN issued a Severe Thunderstorm Warning.

~Gritty

Comparing RAOB to NUCAPS and AllSky Layer Precipitable Water

Davenport launched an 18Z balloon, which gave me the opportunity to compare the RAOB with a NUCAPS sounding (first and second images below, respectively). Initially, I attempted to modify the sounding in NUCAPS to try to bring it closer to the observed values, but after several minutes and attempts at doing that, I realized that I'd have to do multiple levels of modifications before it came anywhere close to the observed sounding. As great as it is to have the ability to modify the NUCAPS sounding, my initial thoughts are that I'm not sure how feasible it would be to do this in a much quicker-paced operational setting. If I'm sitting in the mesoanalyst seat during a severe weather event, I'd need to be able to analyze the available data much faster than doing a more detailed modification would allow.

I was also able to do a PWAT comparison between these two soundings and the AllSky Layered Precip product. The NUCAPS and RAOB are very close together in values, whereas since the AllSky product (last image below) is currently utilizing the GFS to fill in the data in the DVN area, it's noticeably higher (RAOB: 1.0"; NUCAPS: 1.1"; AllSky: 1.3"). I am very happy to be able to underlay the data type for the LAP products, since this is crucial for me to be able to see where the data is coming from and how to correctly assess and apply the right bias adjustments, as necessary.
As for the AllSky Layered Precip product, in general, this is very helpful to be able to identify potential atmospheric rivers and quickly diagnose PWAT trends with a decent degree of confidence, when again combined with the knowledge of what data is being used to compute the output.

~Gritty

Deadly tornadoes in Dixie

Alabama and Georgia were hit particularly hard by a series of tornadoes on March 3rd. A subtle yet energetic shortwave trough with a strong 850mb jet zoomed through the deep south spawning the tornadic storms from the early afternoon through the evening hours. NOAA's Storm Prediction Center (SPC) issued a 10% hatched convective outlook for strong tornadoes (this is 10% probability of a tornado within 25 miles). SPC later issued a mesoscale discussion (MD) for one supercell that formed in central Alabama around 1pm CST (19 UTC). The MD cited pressure falls ahead of the surface low, very strong 0-1km helicity (~500 J/kg), and ample buoyancy and environmental shear for the textual warning of "tornadogensis will likely occur within the next 30-60 minutes with the possibility of a strong tornado occurring."

According to preliminary reports, that was an excellent prediction, as this storm first produced a tornado at about 2pm (at 20:03 UTC). ProbTor, a product of NOAA's ProbSevere*, had increasing probabilities for this storm from 18:36 to 18:50 UTC, jumping from 20% to over 70% in those fourteen minutes (Fig. 1 and Fig. 2), largely due to increasing MRMS azimuthal shear. Strong total lightning density later played a role in increasing the probability of tornado to over 90%. Unfortunately for Lee county, AL, a second tornadic storm followed close behind. The NWS issued a rare tornado emergency at about 20:20 UTC for Lee County, AL and Harris and Muscogee Counties, GA (for the first supercell). These two storms combined for at least 23 deaths in communities between Beauregard and Smith's Station, AL. Many other tornadic storms were reported elsewhere in Alabama, Georgia, and Florida.

Fig. 1: The probability of any severe (contours), with NWS severe weather warnings and MRMS MergedReflectivity composite. The thicker tornado warning polygons (red) denote tornado emergencies.

In AWIPS2, when the probability of any severe product is loaded, a second contour will appear around the radar-identified object when ProbTor exceeds a given value. In Figure 1, we set that threshold to 20%. Users can change this threshold through their ProbSevere USER bundle files. We created this feature so that forecasters can visually see changes in the probability of tornado as well as the probability of any severe. Setting a higher threshold (~10-25%) may be prudent in a strong kinematic environment, whereas a lower threshold (3-5%) may help users visually pick out potential tornadic threats in more mundane environment.

Fig. 2: ProbHail, ProbWind, and ProbTor (red) for the life cycle of this storm. Durations of NWS warnings and times of preliminary LSRs are plotted on the bottom axes.

* The NOAA/CIMSS ProbSevere model (v2.0) will be an operational subsystem within MRMS as early as August 2019.

Outer Banks tornadoes

An energetic shortwave traversing the Ohio River Valley took on a negative tilt and forced several storms on the North Carolina coast. As is somewhat common for the time of year, the kinematic parameters for this environment were very strong -- effective bulk shear ≥ 60 kts, meanwind 1-3km AGL ≥ 40 kts, and storm-relative helicity 0-1km AGL ~200 J/kg.

One storm in particular spawned two tornadoes: an EF0 at 19:06 UTC, and an EF2 from 19:10 to 19:16 UTC, with large sections of roofs removed from homes and powerlines down. The EF2 twister occurred on Emerald Isle.

ProbTornado (a product of NOAA/CIMSS ProbSevere) ramped up quickly to nearly 70% before the first tornado touchdown, but decreased  to 30% as reports came in. In the animation below (Figure 1), you can see the ProbSevere contours (inner contour) and the ProbTor contour (outer contour) colored using the same colorbar (top of image). Increasing rotation evident in MRMS azimuthal shear products, as well as an increase in ENI total lightning density in a very favorable environment led to the soaring ProbTor probabilities.

Fig. 1: ProbSevere contours with MRMS MergedReflectivity and NWS severe weather warnings.

A time series of this storm shows how the ProbWind, ProbHail, and ProbTor probabilities evolved (Figure 2). After a 70mph straight-line wind report at 19:50 UTC, ProbTor increased to 80% when a second tornado warning was issued, but there were no reports. The storm traveled over mostly water at this point, and was starting to reside in a region of high radar beam height, even at the lowest tilt, which potentially may have affected the azimuthal shears.

Nevertheless, this event shows how monitoring the ProbSevere products (namely ProbTor in this case), may help increase situational awareness, forecaster confidence, and ultimately lead-time to severe convective hazards.

October tornadoes

A seasonably potent, negatively tilted trough traversed the central U.S. with an embedded strong 850mb jet, spawning severe weather from Texas to Wisconsin. 

Early in the day, storms surged through central Oklahoma, where NOAA's SPC had a 2% chance of tornadoes within 25 miles of a point (Figure 1). The tornadic nature of these storms was perhaps a bit of a surprise, as a tornado watch was issued at 13:50 UTC, after numerous twisters touched down in the OKC metro area. Structural damage to homes and overturned cars were reported.

Fig. 1: SPC Day 1 tornado outlook at 1300 UTC with tornado reports.

The NOAA/CIMSS ProbSevere model (version2, v2) picked up on this tornado threat shortly before the first tornadoes touched down. In Figure 2, a new AWIPS-2 plug-in demonstrates how the probability of tornado product (ProbTor) is displayed with the probability of any severe. The outer contour is colored by the ProbTor value, whereas the inner contour is colored by the probability of any severe (i.e., the maximum of ProbHail, ProbWind, and ProbTor). Both contours use the same colorbar. The outer contour only appears if ProbTor is ≥ 3% by default, but this can be changed to any value by editing a menu file (note that this feature can also be turned off if the value is set to an impossible value, such as 101%). The developers hope that this outer contour will give forecasters a better visual and quantitative indication for a potential tornadic threat, while still being able to see the overall probability of severe.

In this line of storms, we can see that the two contours appear to have the same color, indicating that the ProbTor value is greater than both ProbHail and ProbWind. Note that by sampling the storm, you can still see the normal readout of probabilities and predictors for ProbSevere v2 (Figure 3). 

Fig. 2: ProbSevere output, with ProbTor contours (outer contours), along with MRMS MergedReflectivity and NWS severe weather warnings. 

Figure 3: ProbSevere with sampling.

Figure 4 shows the rapid increase in ProbTor from 22% at 13:20 UTC to 60% 6 minutes later. This was largely due to increasing MRMS azimuthal shear in an environment characterized by 45 kts of 1-3 km AGL mean windspeed. Interestingly, these storms had no lightning activity.

Fig. 4: Time series for ProbSevere v2 probabilities, with preliminary severe LSRs and NWS severe weather warnings.

In the afternoon, storms spawned tornadoes in Missouri (Figure 5, Figure 6) and Iowa. A storm in Missouri exhibited large fluctuations of ProbTor before producing a tornado. The fluctuations in this case were tied to changes in the MRMS azimuthal shear. Sometimes noisy doppler velocity data contributes to rapid azimuthal shear increases or decreases, so forecasters should always be monitoring base velocity data as well.

In Iowa, numerous tornadoes were reported, but ProbTor was generally < 20% for many of them. Despite a conducive environment, azimuthal shear values were not very high for these storms.

Fig. 5: ProbSevere, MRMS MergedReflectivity, and NWS severe weather warnings for a storm in Missouri.

Fig. 6: Time series of ProbSevere v2 probabilities for the highlighted storm in Figure 5.

No tornadoes were reported in Wisconsin, but there were numerous wind and hail reports. The still image (Figure 7) and time series (Figure 8) show a strong satellite growth rate, moderate MRMS azimuthal shear, and a brisk 1-3 km AGL meanwind (~30 kts) contributing to a ProbWind of 53%, shortly before a barn and power lines were blown down in Iowa Co., WI. ProbWind later increased to over 60% and a tornado warning was issued (ProbTor = 3%), but no reports were received.

Fig. 7: ProbSevere, MRMS MergedReflectivity, and NWS severe weather warnings for storms in southwest Wisconsin.

Fig. 8: Time series of ProbSevere v2 probabilities for the storm in Figure 7, in southwest WI.

If a tornado blows through the Maine woods but no one’s there to see it, did it really happen?

The title of this post is exactly what the Bangor Daily News asked after a supercell snapped trees in the Maine northwoods on August 29th. The average radar beam height of the lowest tilt was over 2 km for much of this storm's lifetime, but the storm exhibited a very strong increase in rotation in the low-levels and mid-levels at around 19:54Z, a few minutes before the first tornado warning was issued. The storm also had markedly weak lightning activity from the ENTLN.

The NWS in Caribou and the Maine Forest Service did a fly over and looked for ground damage but apparently found none. So in this case, it probably didn't happen, but at least any lumberjacks in the area were warned for this potent storm.

Figure 1: ProbTornado contours, MRMS MergedComposite reflectivity, and NWS severe weather warnings. 

Figure 2: Time series of the probability of tornado and constituent predictors for the tornado-warned storm in Figure 1.