Week 1 Wrap Up

The first week of the 2018 HWT Satellite Proving Ground Experiment is wrapping up today. It has been a great week with active weather and 5 forecasters providing some valuable feedback for the week on a host of products. Monday will begin week 2 of the experiment.

Michael

Tornadic storm near Des Moines

As a surface low pressure system tracked northeast through central Iowa last night, one storm near the center of the low spawned at least one tornado southeast of Des Moines, IA. The first report came at 00:38 UTC, about 13 minutes after the first tornado warning. The ProbTor product from ProbSevere version2 showed elevated probability of tornado values (≥ 20%) as early as 23:48 UTC.

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

From the time series below, we see that increased rotation (0-2km MRMS AzShear; solid orange line) and increased total lightning density (solid green line) helped jump the probability of tornado from single digits to 40% at 00:00 UTC, and again from 20% to 50% between 00:26 and 00:38 UTC. The effective bulk shear and 0-1 km storm-relative helicity were clearly adequate, but not outrageous (~40 kts and ~150 m2/s2, respectively).

Figure 2: Time series of ProbTor and predictor values. The bottom axes contain NWS warning durations and LSRs.

One thing forecasters should look for when using ProbTor is when a storm's probability of tornado "sticks out" amongst the values of neighboring storms. This storm is a good example of that, where neighboring storms in a 4-county radius had probabilities in the single digits.

Forecasters should also be aware that the probability will fluctuate more with ProbTor than the probability of severe. This storm exhibited some fluctuation, whereby probabilities rose from 7% to 43%, dipped to 20%, and rose again to 55% during the tornado(es). ProbSevere developers are working on incorporating a visual time series function in AWIPS2 for ProbSevere storm objects (similar to Figure 2), so that forecasters can quickly see the history of a storm, which may help in warning decision making.

Lastly, there is no "magic" threshold of ProbTor for when tornadogenesis is imminent. Very potent environments characterized by SPC's "moderate" or "high" risk for tornado might see ProbTor values in the 70-90% range for tornadic storms, whereas for a more common tornadic environment like this, 20-50% can be expected for tornadic storms. In environments where tornadoes are not expected, ProbTor values of even 10% might be significant.

Users of the ProbTor product should keep these points in mind, but ultimately, forecasters should become more comfortable with the product as they gain experience with it in different regimes.

Please see our training module for more details on ProbHail, ProbWind, and ProbTor, and this fact sheet regarding the differences between ProbSevere version1 and version2.

GLM 1-min vs 5-min Accumulation

For most of the week, the products from GLM were 1-minute updating every 1-minute. For this user, it was difficult to identify relevant trends in the fields (rapid increases or decreases), or to connect lightning cores with their appropriate storms as seen in radar imagery. Today, the products were converted to 5-minute accumulation (updating every 1-minute) about midway through the shift, allowing for a comparison between the two methods. See below. The 5-min accumulation updating every 1-minute, in my opinion, allows for much easier identification of significant trends, as well as tracking of lightning cores and relating lightning features to radar imagery.

The animation below includes 1-min visible imagery with 1-min GLM Flash Extent Density overlaid. The loop begins with the 1-min product, and transitions to the 5-min accumulation product updating every 1-minute.

The loop below includes 5-min visible imagery with the 1-min product overlaid, transitioning to the 5-min product midway through.

- Bucky

Give me Flash Area or give me death - P Henry

A good IDSS example.

A single product such as Flash Extent Density provides limited information about a flash and potentially storm behavior. Pairing FED with Flash Area as in this example helps more easily diagnose the long flash through the stratiform region of this storm over northern IA. The higher flash rates associated with the smaller flash sizes along the leading edge of the northward moving complex are more easily explained with the companion product. As the flash extended north into MN and another CWA, this information could be used to enhance DSS since the main convective line was still 50-60 mi away. ENI and other commercial lightning networks would not have alerted to the threat of lightning overhead at such a great distance. An event organizer or EM looking at static radar also may not be aware of the increased threat of lightning well ahead of the main storm band. SCoulomb

ProbSevere running hot in Iowa

ProbSevere has been running hot, with probability of hail and probability of wind, pulsing into the 30-50% and 60-80%, respectively for many storms across the DMX WFO.  Thus far no severe reports have been received.  In discussing the environment with the DMX team, we are wondering if the shear is too high for the amount of instability, which is shearing the tops off developing storms, preventing storm organization.  The tops being sheared off can be seen in 1 minute GOES-16 visible imagery (not shown).  One silver lining would be ProbSevere Version 2 probabilities are 10-40% lower than the ProbSevere Version 1--illustrating the better calibration of probabilities in ProbSevere Version 2 relative to Version 1 (although still too high for the small sample size of a few hours this afternoon in the DMX region).

-J. Sieglaff

CI is Trying It's Best

Here is a 2-hour VIS loop of 1-minute data.  Overlayed is CI Probabilities.  The values seem to bounce around a lot, but are overall pretty low.  Even in locations where they are higher (green and yellow) not much convection develops from the cumulus.  The cirrus contamination diminished the potential.  Overall, the product needs more work to improve forecaster confidence.

-Kevin

Artifact in GLM near 26N

There is a known artifact in the FLM data.  Note the linear feature north of the yellow line (ZMA - Miami CWSU boundary) in the upper left and lower right.  This is near 26N and will always appear here, as it is some boundary in the gridded data that's stitched together (as I understood).  GLM development team is aware and working on getting rid of it.  In the example below there was some convection to the north of the artifact.

-Forrest

NUCAPS Across Iowa and Surrounding Areas

Modified NUCAPS came in just as convection began to initiate today. Repeating what I said yesterday, I appreciate the update Jack made to adjust the surface values of all "green/passed QC" NUCAPS points using RTMA, even if the more intensive adjustment cannot be made. The failure of the more intensive adjustment for passed qc soundings is due to a lack of GOES surface temperature data due to low cloud cover.

Sampling NUCAPS soundings from the central portion of our CWA down through south of our CWA, you can see the trend of increasing moisture and instability. This trend is observed in the three soundings below. Looking at SBCAPE, values less than 500 j/kg in central Iowa increase to around 1500 j/kg in southern Iowa to 2300 j/kg. This general pattern verifies what SPC mesoanalysis is showing, and tells us that instability is increasing from the south to values that will support strong updrafts and development of severe weather.

Soundings in the area of early storm development indicate the -20C level around 20,500 ft. High reflectivity cores reaching this level will be relevant given this will be a preferred hail growth zone.

 

1

 2

 3

- Bucky

FED vs GED (Event Density)

Over the course of the week I have been concerned with comparing the consistently low FED values (single digit for most storms) with the other lightning fields including ground-based networks. Visually the low counts per pixel do not grab your attention when looking at a busy scene with storms in a variety of stages of their life cycle. The PIs suggested using Event Density (group extent density) as an alternative with the understanding that it will show higher values. To me the GED compares more nicely with the other GLM products than the FED and also helps in distinguishing core intensity within a storm complex.  It also compares more favorably with the number of events from ENI data for easier transition from one network to another. The example below all shows the striping across Pennsylvania as the pixel sizes are adjusted further away from nader. - SCoulomb

GLM and Convective Inititation

Saw an interesting example of a developing CB (over land at the Haiti/Dominican Republic border) where GLM flashed before development was recognizable on satellite.

First, a longer loop.  Then a shorter loop focused on three frames.  All data is time matched to satellite.  In addition to GLM, ENI is plotted on the upper left (small yellow dots).

Satellite/Lightning 2002-2057Z

At the center of each of the four images is the developing cell.  The first flash occurs on frame 4 of 12.  It's most noticeable on the lower right with VIS and Total Energy.  Then, notice on the upper right the enhanced IR.  The flash seemed to practically come from nowhere, but then the cloud becomes more recognizably convective a couple frames later.

Three framed loop.  Initial flash is frame 2.

Satellite times above are 2012-2017-2022Z, with the initial flash at 2017Z.  Again, notice the upper right enhanced IR with little to see.  Max tops estimated by cloud top temp in the sequence was FL280-FL320 (flash)-FL350.  The cloud top ended up a frame or two later peaking at FL380.  ENI didn't pick this up for a few frames after this loop.

Sequence of GLM product values for the event:

Total Energy:  0fJ - 52 - 0.4
Flash Extent Density:  0 - 1 - 1
Event Density:  0 - 42 - 3
Area Extent.   0km^2 - 1122 - 70.2

Eric happened by while I was checking this out and made an interesting comment.  The intense, single flash was likely seen very clearly from the satellite since the icing/charge must have been restricted to the cloud top.  Thus, GLM can be useful for recognizing convective initiation.

-Forrest

GLM Imitates a Zebra

Notice the GLM data.  At first it appears to be "missing" or "broken".  In actuality, this is an artifact of the location.  The resolution of GLM changes as one gets further away from nadir.  One of lines where the resolution changes happens to be over PA, where this image was taken.

-Kevin

Convective Initiation Product for Des Moines, IA

Convective Initiation (severe) product indicating developing TS across far southern IA late this afternoon. The values indicated look to be running too 'hot' (Figure 1). Base reflectivity 5-10 minutes later verifies what the CI product indicated (Figure 2).

 

Figure 1. Convective Initiation Severe values increasing across far southern IA and far northern MO late this afternoon.

Figure 2. 4 panel of ProbSevere products overlaid with 0.5 degree reflectivity. Top Right: ProbHail; Bottom Right: ProbTor; Bottom Left: ProbWind

NUCAPS Gridded Low-Mid Level T & Td

One of the forecast challenges in Florida is detecting "blobs" of moisture advecting west from the open Atlantic/Bahamas into the peninsula.  These areas of increased moisture can indicate a wind surge, and increase nocturnal showers, or provide additional moisture for increased coverage of sea breeze thunderstorms during the day.

Having additional ways of identifying these features is always welcome.  In addition to the NUCAPS images of H7 and H85 moisture, I've included temperature, and compared with RAP.

NUCAPS H7 Td

RAP H7 Td

NUCAPS H85 Td

RAP H85 Td

At 850mb, prevailing flow today is from the east.  On the RAP and NUCAPS images above at this level, you can see elongated (white-red extending N-S, such as over S FL and over the eastern gulf) areas of moisture, as described above, moving westward.

NUCAPS H7 T

RAP H7 T

NUCAPS H85 T

RAP H85 T

When evaluation NUCAPS temperatures, there is good agreement between NUCAPS and RAP when look at relative values.  This seems to be the better interpretation of NUCAPS data that I've found so far, rather than looking at absolute temps.  We can see from the smoothed out point soundings that literal values are not it's strength.  Rather, "Temps at this level are higher over here, than they are over there" is more NUCAPS strength.  That still has value in situational awareness.

The satellite scene 1830-1915Z during time of NUCAPS data

The satellite scene has some patches of lower level clouds over FL and cstl waters.  There is some cirrus streaming SSE over the eastern gulf.

-Forrest

NUCAPS crosses the Delaware

A look at the environment across the BGM area ahead of the approaching convection confirmed weak instability with MUCAPE <1000 J/kg on the RAP forecast sounding near Mt. Vernon. A NUCAPS sounding in the same location an hour earlier depicted greater available instability (almost 1500 J/kg). Boundary layer temperatures were similar on both but NUCAPS showed a more moist profile. The big difference was the warm layer above 500 mb depicted on the RAP sounding and nearby ALY 12Z run. NUCAPS did not have this warm layer and therefore higher overall CAPE values. - SCoulomb

Mesoscale Analysis in Des Moines

GOES-16 upper-level water vapor imagery reveals a closed upper low advancing east across southern Nebraska early this afternoon.

Visible imagery shows low clouds across much of the CWA, with cu beginning to develop across the far southern portions in the warm sector of the storm system

Morning GOES-16 Derived Motion Winds revealed a 110+ knot 300 mb southwest to northeast oriented jet core across Iowa. This agrees with the GFS depiction of the jet in that area, which has that jet exiting and a jet entrance region poking into southern Iowa later today. Winds in the 750 mb to 850 mb layer are generally 15-25 knots from the south/southeast, with slightly stronger winds advancing towards our south implying increasing LL WAA.

The default LAP products have very little data given low cloud cover over the region.  However, the all-sky LAP products have a lot of cloudy sky retrieval pixels, filling in those gaps, in addition to some clear sky pixels moving north into the CWA. From 1730 to 1900, all sky LAP CAPE has increased from 500 to 700 j/kg, and TPW has increased from around 1.32" to 1.36". CAPE over 1000 j/kg reside just south of our warning area. These trends imply the atmosphere is becoming more favorable for the development of severe storms.

 The LAP Layer PW product, comprised mostly of clear sky and cloudy sky retrievals across the domain, shows some significant features. Clockwise from top left PW sigma layers: sfc - 0.9, 0.9 - 0.7, TPW, 0.7 - 0.3. The two lower layers show fairly high levels of moisture, while the upper level reveals a nose of dryer air advancing toward the CWA. This trend would imply increasing convective instability as we advance through the afternoon.

- Bucky

GLM missed low energy/first flash

A couple of times this week we have noticed in comparison with ENI, missed flashes on GLM. The first flash is typically the most important for IDSS for outdoor venues, especially since the need for lead time grows with increased size of crowds. In order to get the most lead time there likely needs to be a trade-off with higher false detections for GLM to keep up with ground-based networks for the weaker events. It is preferred that GLM has equal or better detection capability for low energy and first flash as the ground-based networks in order for it to be trusted for application in IDSS. The more times flashes are depicted by other sources and not by GLM the less GLM is going to be trusted in a forecast environment despite its other benefits.  Similarly the more ECMWF data shows a faster fropa with arctic outbreaks the less the GFS is going to be used to time shallow fronts.

[no image example]

ProbWind Underdone and Missed a Warning

Here is a 50 minute loop of 0.5 Velocity data from BGM.  Notice the strong area of winds approaching the radar.  It's around 50 kt a few thousand feet off the surface.  The second image measures 52 kt at 2,500 feet AGL.  But ProbWind reported 1% probability for virtually the entire time.  This is massively underdone.  Why?  50 kt is Severe Thunderstorm Warning criteria.  If this wind a few thousand feet above ground mixes down to the surface, then it's easily SVR criteria.  Yet the ProbWind is only 1%.

So what happened?  At 1927Z (image three) NWS BGM reported tree limbs and wires down on Timberline Drive.  This matches up fairly well with the general location highlighted in image two.  If forecasters were relying solely on ProbWind, they would have missed a warning.  It needs to factor the SRM and Velocity data into the calculations.  Otherwise, forecasters may not be as confident in the ProbWind data.

 

-Kevin

PW's/Instability Increasing across DMX (Des Moines, IA)

AllSkyLAP PW product indicates moistening in the low and mid levels across DMX FA this afternoon (Figure 1). AllSkyLAP Instability parameters are also on the rise (Figure 2).

Figure 1. AllSkyLAP PW product. Top Left: SFC-900mb; Top Right: 900mb-700mb; Bottom Right: Layer PW; Bottom Left 700mb-300mb

Figure 2. AllSkyLAP products. Top Left: Layer CAPE; Top Right: Layer LI; Bottom Right: Layer Cloud Type; Blue=Clear Skies; Gray=GFS First Guess; Yellow=Cloudy Skies

- BO

BGM initial thoughts

A more subtle severe weather threat is expected relative to the past couple of days. Convection to the west is spreading east into the BGM area at the start of operations. Overall instability is weak with observed and SPC analysis CAPE ~500-1000 J/kg at best. Shear is strong and storm motions may be on the order of 50kt owing to a severe wind threat. The southern half of the BGM area seems most at risk for severe weather with better instability south of a boundary slowly sinking south with time.

ProbSevere Provides Valuable Lead Time for ICT

ProbSevere provided valuable lead time on redevelopment of MCS in northern portions of ICT CWA after 6:00pm local time. Quarter size hail was reported in Madison County (two counties NNE of radar) at 6:15pm, where ProbHail increased to over 80% (Figure 1). ProbWind remained rather high through the event (>80%), so was less useful, but became maximized as the storm moved through Chase and Greenwood counties, where wind gusts of 65-70 mph were reported.

Figure 1. ProbHail spiked back up on storm over central Marion County (two counties NNE of Radar) between 6:00pm and 6:15pm when quarter size hail 

was reported.

Figure 2. ProbWind proved less usefull as values remained quite high 

throughout event, but severe wind gusts were reported in Chase and northern

Greenwood counties. 

- BO

Week 1 Day 4 Operations

For today we will be beginning operations in the Des Moines, IA and Bighampton, NY CWAs.

Michael

Week 1 Day 3 Wrap Up and Feedback

We had another busy day yesterday with a tornado warning over us during operations! Feedback from the day is posted below:

ABI and All-Sky LAP

CAPE as usual seemed to be lower than it should
Where there was clear air it seemed to get a good retrieval but the area where the GFS retrieval is
taking over is scattered and pixely

Shows a blank screen every hour on the hour when there is no retrieval

DMWs analyzed the jet well with the diffluence aloft where the convection was firing

Could see the lower level winds start to back as the evening wore on just as surface winds started to
back

Can’t really use the day cloud phase for monitoring storm top features since the Vis gets washed out

Getting RGB application labels built into AWIPS would be nice to mouse over for labeling of what is
represented in the RGB similar to the radar Hydrometeor classification.

Would prefer to see the actual readout from the channels on RGB instead of a number from 0-255

CI

Probabilities ramped up all over the place really but wasn’t specific to the location

Seemed overly sensitive and jumped around a lot

Wasn’t used a lot due to a couple of known issues being worked

NUCAPS

A fix was put in that every green point will at least have a surface moderation if the rest of the algorithm
doesn’t modify the profile

Gridded product hasn’t gotten much use due to a large number of clouds in the area throughout the day

Compared the profiles to the special Norman sounding and it didn’t pick up the inversion or the EML as
well as what was present

700mb temps seemed to match pretty well compared to RAP analysis to monitor CAP strength across
the area

ProbSevere

Preference for the all hazards and for ProbTor to be on the same display

ProbWind seemed to be high (80 - 90%) across much of the time when the line was sub severe

Some kind of velocity information would be useful for wind information

Probabilities lagged slightly behind the base data analysis for a couple of storms in Kansas

Might be useful to highlight areas of maximum Az shear in line segments after reaching a certain
thresholds based on distance from radar

As storms were splitting the contour would highlight both storms and cause confidence to be lost for
those splitting storms

GLM

It was difficult to determine which lightning cores correlated to which storms in the radar

It would be better to have a cumulative 5 minute product that updates every minute to better see trends
and relate to the stronger cores

It was helpful to have the flash area and flash energy to help pick out the storms different behavior and
distinguish the stronger updrafts from weaker updrafts

It would be most useful to be able to get these products into one product to be able to be used by a
forecaster

Having useful examples for training will be helpful

Michael

GLM Loop

Here is a 30-minute loop of GLM data.
Top Left is Flash Extent Density
Top Right as Average Flash Area
Bottom Right is Total Energy
Bottom Left is 0.5 Reflectivity

Notice the increase in lightning around the perimeter at 00:18Z.  This is probably anvil lightning being detected across the entire line.

-Kevin

God help us...he's trying to issue warnings!

Had fun pretending to issue warnings today.  Was monitoring a cell that had a persistent, not super impressive mid level rotation.  Suddenly the cell developed a very strong low level rotation on the 0.5 tilt of the FDR radar.  I definitely wasn't expecting to see something develop such a strong signal that would immediately make me think of issuing a TOR.  I easily could have been paying attention to some other cells.

At the time that the low level circulation became visibly obvious to me, ProbSevere had 17% for TOR (values above 80% for wind/hail like most storms today).  In one or two updates it jumped to 72%.  ProbSevere showed potential to draw my attention back to the cell very shortly after development had I been looking elsewhere.

(FDR radar SRM loop)

-Forrest

SNPP NUCAPS Near Norman

SNPP NUCAPS valid shortly after 1910 UTC was available in AWUPS around 2000 UTC, providing an opportunity to compare with the 1800 UTC Norman radiosonde. Jack adjusted the "Modified NUCAPS" code to allow for surface modifications (using RTMA) still even when the more intensive modification cannot be made. This worked, and prevents the user from having to modify the the surface. The shape of the NUCAPS profile was generally on par, missing the temperature inversion, not surprisingly. The SB LCL was very similar to the raob, but SBCAPE values were around 500 j/kg too high.

- Bucky

NUCAPS plan view shows warm mid levels

Near the time of the NUCAPS pass, warm mid level temps, above +10C bulged northward from Texas to the eastern two-thirds of OK according to regional RAP analysis.  Plotting of NUCAPS H7 temps revealed the warmest temperatures in the same area, though the warmest temps were a little cooler, 7-8C.  Temperatures were cooler to the east of the state and in the panhandles.

(NUCAPS H7 temps on 05/02)

-Forrest