Week 1 (May 4-7 2015) Summary and Feedback

During the first week of the 2015 EWP Spring Experiment, we had forecasters from the San Diego, Eureka, Huntsville, Sterling and Norman WFO's, as well as a broadcast meteorologist from NBC4 Columbus, OH participate in the Spring Warning Project. With most of the active weather confined to the Southern Plains throughout the week, our CWA's of operations included: Topeka, Omaha, Wichita, Midland, Lubbock, Amarillo, Albuquerque, San Angelo, Norman and Hastings. We had a good mix of marginal and very busy severe weather days, with Wednesday being the most active weather day as severe weather broke out across central Oklahoma.

Participants were able to use all of the demonstration products this week, which included GOES-R and ENI total lightning products. There were many good blog posts written throughout the week highlighting the use of all of these products in various forecast/nowcast/warning situations.

Below is a feedback summary from the GOES-R demonstrations during the first week of the 2015 HWT Spring Experiment, collected at the week 1 debrief.

PGLM
- It was useful yesterday in the Lubbock CWA.
- I have a marine responsibility in my CWA, so the lightning data would be useful for issuing sub-severe forecast products for storms moving towards the coast.
- This lightning information would be useful in my CWA on days when the fire danger is heightened.

Lightning Jump
- I was not so sure if a 1 or 2 sigma jump was significant, or if I should wait for a 5 or 6 jump.
- At one point when things were active, I was ignoring the 1 sigma jumps. The 4 and 5 sigma jumps really drew my attention.
- 3-sigma was when I really started paying attention to the storm.
- 1-sigma is probably not even worth a color, I made it transparent.

ProbSevere
- It did a good job with discrete cells, but when the mode became more linear, it suffered
- It really drew my attentions to the storms that I should interrogate further
- For me, MESH in the readout does not need to read to hundredths of an inch.
- Color/highlight values in the text display as they become more significant. Make the actual probability stand out more too.
- For the color contour, I made it neon on the higher end, and got rid of the lower end. The low end looked almost the same as the higher end.

CI
- Overall it did a pretty good job throughout the week of highlighting where CI would occur
- It really works best in a clear environment (no cirrus contamination)
- When you had a cu field developing, it did a nice job of depicting where stuff would go
- I liked using it when we had a lot of boundaries, it did a good job of depicting where along the boundaries stuff would go. Especially today when there were  a lot of different boundaries in play, it gave me confidence where convection would develop and where I should look.
- Having a higher threshold for CI would be more relevant in WFO operations, and less messy
- With the colortable, the lowest probs (deep blues) stood out the most, which is not what you want to see. The higher end colors against the light background were more difficult to see. Reduce the appearance of the lower probs make the higher probs stand out more.
- Perhaps you could start out with white to light gray at the low end, and transition to colors at the higher end.

GOES-R LAP
- I use the GOES PW field the most
- I would like to know what data are from the retrievals and what are from the GFS.
- The pixelation in certain areas was an issue, with sharp, unnatural transitions present between adjacent values.

NUCAPS
- It might be a good idea to merge this product with something like LAPS to improve the lower levels.
- I want to set this up as soon as I return to my office.
- I can see myself using this a lot in the winter.
- I esepcially like the observation-based nature of it
- I would like to look at this over Lake Superiror
- Anything that gives us a temperature profile will be helpful, especially in the winter.
- In San Diego, it will benefit us during the summer monsoon. Also, the San Diego RAOB is not representative of the mountains in our CWA
- QC flags would give me more confidence in the soundings, as it is difficult to judge with just the cloud data.
- The RAOBS in my area are not representative of most of my counties, so I often use forecast soundings.

- Bill Line, SPC/HWT Satellite Liaison

Daily Summary: Week 1 Day 4 (May 7, 2015)

Today's operations were in the Norman, Amarillo and Lubbock CWA's. Compared to yesterday, it was a much quieter day, and convection took a little longer to get going. This allowed forecasters more opportunity to evaluate environmental analysis tools such as the GOES-R LAP algorithm and JPSS NUCAPS soundings. The GOES-R CI algorithm was also utilized by participants in the pre-convective environment. The Lubbock pair had the opportunity to evaluate the PGLM total lightning, especially late in the day when convective activity amped up in intensity and coverage. With the Frederick, OK radar down for most of the day, the Norman group had increased reliance on the Earth Networks lightning tools.

Tomorrow we will have our weekly debrief, and participants will complete their end of the week surveys and present the Tales from the Testbed webinar.

See Week 1 Summary for more feedback from this week.

- Bill Line, SPC/HWT Satellite Liaison and Week 1 EWP Coordinator

Convective Initiation and Radar

With the OUN cwa quieting down, decided to take a look at how the convective initiation product was working this afternoon. The animated image below (click to view), highlights a couple areas (both in the center and upper left of the image) where the CI algorithm picked up on the development of showers and thunderstorm. Meanwhile, in the upper right hand part of the image, there were several instances where the CI algorithm showed probabilities over 60% and nothing developed as of 2245Z.

-SRF

Lightning Data Prompts SVS

LBB remains a convective mess. Overall storms have been sub-severe. The flash extent density for a cluster of storms in the extreme SE portion of the CWA caught  my eye.

The two light pink pixels were around 56-57 (50 being the general minimum for severe storms) at 2153Z.

Looking at the other guidance we could use today, there was a 3 sigma lightning jump and a prob severe of 90% at 2149Z.

I went ahead and issued the warning based on the lightning and prob severe guidance, despite the storms not looking overly impressive on radar. As has been the case the rest of the day across our area, there have been no storm reports.

-V. Darkbloom

Comparing storm development with satellite derived CAPE/LI

Used satellite LAP algorithm to track the change in the environment over the course of the afternoon. Best instability was in the southeast CWA. Overlayed one hour total lightning with the CAPES and LI to see how the coverage of storms lined up with the best instability. Image shows this overlay at 21z.

 Ongoing MCS from this morning accounted for the lightning east and south of the forecast area. Narrow strip of lightning west of the better instability was where outflow boundary intersected cold front/dry line. This area had better low level lapse rates and more time to recover after the morning MCS.

The mesoanalysis from SPC showed much higher CAPS at 21 and 22z than the values from the LAP algorithm. SPC was 1000-1500 J/KG while satellite derived was less than 500 J/KG.

Lynford

ENI Time Series and ProbSevere

New convection has been initiating along the nose of stronger instability feeding northward across north central TX and I have been watching one cell in particular in Young county TX.

The ENI time series has provided a nice trend in the strengthening of the storm and showed a large jump in flash rate just before 2050 UTC.  Looking at the base data all tilts at the time of the jump revealed a deep 50 dbz core to almost 40kft and a 70 dbz core to over 21kft.

ENI Time Series

I think this ETN time series data was useful in identifying a rapidly developing updraft. Interesting to note, the ProbSevere model indicated a 74% severe threat at 2024 UTC on this storm which then went to 88% at 2033 UTC and then 98% at 2040 UTC.

Radar image loop with ProbSevere overlay

The ProbSevere output on the storm when it showed the 74% threat showed a strong glaciation rate and moderate growth rate with MESH of 0.68 inches.

Jack Bauer

NUCAPS in Amarillo CWA

Three NUCAPS soundings were sampled this afternoon around the Amarillo CWA.  Several boundaries are observed in the CWA today…of most interest, a westward propagating outflow boundary from morning MCS and the dryline.  The northern and central soundings were west of the ouflow boundary, but east of the dryline.  Both of these soundings were about 8 degrees too cool (temperature) and 15 degrees too dry (dewpoint) at the surface.  Once the profiles were modified, they yielded 1500 J/kg and 2770 J/kg CAPE, respectively.  The sounding behind the dryline was a close match to observations…and little modification to the sounding was needed.

Blue line is westward moving outflow…dryline is marked in yellow.  NUCAPS soundings modified are circled in black.

ALEXANDER’S DARK BAND/LYNFORD

EWP AMA team issues experimental severe thunderstorm warning

The forecasters covering the AMA CWA issued an experimental severe thunderstorm warning for a thunderstorm over the Oklahoma panhandle at 2114 UTC 07 May 2015.  The NOAA/CIMSS ProbSevere value jumped to 87% for this storm at 2110 UTC 07 May 2015 (Figure 1).

Figure 1.  NOAA/CIMSS ProbSevere and MRMS composite reflectivity valid at 2110 UTC 07 May 2015.

The forecasters attributed their warning to their assessment of the environment, HRRR forecasts indicating potential for sustained updrafts, NOAA/CIMSS ProbSevere trends, and analysis of base radar data.

UPDATE:  2136 UTC 07 May 2015 the AMA team issued another experimental severe thunderstorm warning for a storm in Oldham County, Texas.  The team noted high ProbSevere values despite the MRMS MESH being just under 1.00".  Despite the sub-severe MESH, the ProbSevere higher probabilities (80+%) are attributed to moderate derived vertical growth rates and glaciation rates and a favorable environment in addition to near-severe MESH values.

-Sieglaff

Convective Initiation Miss

At 1930 UTC the Convective Initiation product showed a probability over 60% over Montague/Cooke/Denton counties in TX.

However radar imagery over the next hour did not show any development with reflectivities 35 dbz or higher, in fact very little at all was noted.  The image below shows the radar data at 2030 UTC with little if any returns.

Jack Bauer

pGLM, lightning jump, and prob severe

Amongst a mess of convection in the Lubbock CWA, some of the new tools definitely help focus your attention on certain cells.

One sigma lightning jump

Just looking at the radar data, not much stands out. However, the lightning jump and prob severe (86%) definitely catch your attention.

At 2016Z, the lightning density peaked at 81, the highest I’ve seen so far.

By this time the prob severe was up to 94%. Based on radar data the storm looked like it intensified for a short time and was definitely stronger than other cells in the CWA. Nothing was reported with it (so far). This is a good case for using these tools for situational awareness, and perhaps giving you longer lead time on certain cells within a mult-cell cluster or line of convection.

-V. Darkbloom

Strong satellite growth lends great lead-time in southern Nebraska

Convection was erupting along a cold front over portions of Nebraska and Kansas Thursday afternoon.  A storm along the Nebraska/Kansas border caught our attention.  The NOAA/CIMSS ProbSevere model had a probability of 10% at 1932 UTC 07 May 2015 (Figure 1).  At 1934 UTC the latest GOES-dervied growth rates (both strong vertical growth and glaciation rates) made it into the ProbSevere model and the probability increased to 43%.  Four minutes later at 1938 UTC, the radar observed intensity continued to increase (MRMS MESH of 0.57") and the ProbSevere output jumped to 69%.  At 1940 UTC, the MESH sized continued to increase (just over 1.00") and the ProbSevere probability jumped to 96%.

Figure 1.  NOAA/CIMSS ProbSevere and MRMS composite reflectivity valid 1932 - 1940 UTC 07 May 2015 centered on the Nebraska/Kansas border.

This storm went on produce severe hail (1.25") at 2004 UTC--24 minutes after the ProbSevere output jumped to 96%!  This example showcases how the different ingredients of the ProbSevere model work together to produce a skillful forecast with good lead-time.   At 1934 UTC, the strong derived vertical growth rates and glaciation rates caused a significant jump probabilities, but the radar estimated hail size of only just over 0.50" acted as a restraint--keeping the probabilities from being too large too quickly.  As soon as the radar estimated hail sized increased closer to severe thresholds, combined with environmental information and the GOES-dervied growth rates the probabilities jumped into a very confident range (in the 90% range).

-Sieglaff

UAH/GOES-R CI across southern LUB

The predominant scenario for convection across the central/northern LUB CWA is the continued slow movement of the MCS. A weak outflow/remnant boundary exists west and south of the slowly easterly moving MCS that has acted as a trigger from some storms to fire as the mean southerly flow advects a shallow cumulus field toward the MCS. The dryline remains just west of the CWA and may be a focus for storms later this afternoon.

The CI field below continues to possess high percentages as it advects towards the outflow/remnant boundary. Overall it has been a good situational awareness aid. The example below shows a small cluster of cumulus with higher percentages near Girard, TX, at 2007Z that crossed the outflow boundary and quickly ramped up with increasing reflectivity values well above 35 dBZ (images from 2003 to 2011Z).

Brick Tamland

Currently Using ProbSvr for Situational Awareness

Storm in the southern portion of our forecast area have been posing the greatest threat for severe weather, actually receiving a 2 inch hail report near Mabelle TX.  While focusing my attention there, I have been using the ProbSevere model to provide me with general situational awareness for the north-south line of storms further to the north.  I will also run through the all tilts base data just to confirm there is no significant threat to monitor but this is a nice tool to have to provide situational awareness for multiple storms in the forecast area, especially if you are in a situation where staffing would not allow for sectorization.  Notice in the image below the very low ProbSevere values along the north-south line of storms.

Jack Bauer

GOES CI in Amarillo CWA

Morning MCS over the southeastern Texas panhandle has resulted in a north-westward propagating outflow boundary.  This boundary has merged with a loosely defined dryline in northern parts of the CWA, with the dryline then extending further to the south.  More elevated values of GOES CI (30-60%) has focused very well along these boundaries.  An area of 75% probability was indicated at 1937 UTC.  By 1956, a weak echo was apparent in the AMA radar reflectivity.  By 2008 UTC, the intensity had increased to 50-60 dBz.

ALEXANDERS DARK BAND/LYNFORD

LBB Prob Severe and Lightning Jump

Looking at prob severe and the lightning jump associated with a particularly nasty looking storm.

 The prob severe at 1918Z – 52%


1924Z – prob severe jumps to 70%


1926Z prob svr – up to 85% with a 3 sigma lightning jump


1927Z – 65 dBZ at almost 27 kft AGL!

A severe thunderstorm warning was issued at 1932Z. No reports yet.



-V. Darkbloom

Storms Firing Along Outflow Boundary in Southern OUN CWA

Strong thunderstorms are developing over northwestern TX along an outflow boundary left over from convection over OK and northern TX.  The ProbSevere values increased between 1902 and 1916 UTC (Figure 1).

  Figue 1.  NOAA/CIMSS ProbSevere Model, composite radar reflectivity and GOES-EAST visible imagery valid 1916 UTC 07 May 2015.

The Norman forecasters have noted the uptick in the ProbSevere data with these storms.  The ProbSevere data combined with analysis of base radar data and knowledge of the storm environment was used to issue experimental severe thunderstorm warnings at 1921 UTC (northern warning)  and 1924 UTC (southern warning) (Figure 2).  The ProbSevere probabilities first exceeded 50% for storms A-D in Figure 1 at 1904 UTC, 1926 UTC, 1910 UTC, and 1914 UTC, respectively.

Figure 2.  NOAA/CIMSS ProbSevere and KDYX 0.5 degree reflectivity valid at 1926 UTC 07 May 2015

-Sieglaff

Warning issued for Storms over SW OUN CWA

Rapidly intensifying storms entering the OUN cwa where picked up well on ProbSevere, with them quickly rising to 90%. Based of the reflectivity cores aloft, broad rotation, and the rapid increase in ProbSevere, have issued warnings for multiple storms in the SW corner of the cwa (at 1922 and 1924Z).

While the ProbSevere and cores aloft support the warnings, lightning data has been struggling so far with the peak lightning cell data only around the mid 20s.

Finally, here is a look at the time series for the strongest cell, with a few jumps but overall limited flash amounts.

Will be monitoring the lightning trends and see if the data catches up with the radar and ProbSevere appearance.

-SRF

Comparison of ProbSevere and Custom DTA

After setting up the custom Thunderstorm Alerts (75/min and 100/min), there were a couple areas of focus on the 100/min alerts (to the west-southwest and southwest of KTLX).

But when comparing to the ProbSevere data, we are seeing different representations on the strength of the storms. The storm to the WSW only had a ProbSevere around 11% (that had the greatest flash rate around 149 per min), while the storm to the SW had a ProbSevere of 88%, but a weaker flash rate (87 per min).

These differences highlights the need for analyzing multiple data sources and comparing to radar.

-SRF

Lubbock Today – Initial Look

This morning iin the TX panhandle there was an existing west-east line of thunderstorms stretched across the northern portion of the LBB CWA. A warm front was aligned SW-NE across the AMA CWA, with a dryline across the southern panhandle and a nice dryline bulge south of the CWA.

GOES-R PW


GOES-R CAPE

CI has tended to be more of a “now”cast as opposed to a forecast.



CI at 1815Z


Radar at 1814Z


Radar at 1824Z


CI at 1825Z

By the time the CI was high, convection was already ongoing in that area. More frequent satellite data would definitely help in these rapid development situations!

CI ahead of shower development

CI from 1825z (image 1) showed an 83% probability of development. Showers first reached 35+ dBz at 1843Z, 18 minutes later over Texas County, OK (image 2).

Lynford

Week 1 Final Day

The forecast crews are setup in Norman, Lubbock, Amarillo WFOs today for the final day of the week 1 EWP operations (Tales from the Testbed will occur on Friday).

The Norman crew has a lot of ongoing convection in the southwestern portion of the CWA.  While they were getting acquainted with the setup and datasets for the day one forecaster noticed a rapid jump in the NOAA/CIMSS ProbSevere output on a cell ahead of the bulk of the ongoing convection, the forecaster's blog post is shared below:

Rapid Increase in ProbSevere:

While we were writing up a previous blog post, we noticed rapid increase in the ProbSevere for a storm to the west of the OUN radar. The ProbSevere quickly ramped up from 9% to 94% as seen in the two images below.

Figure 1:  NOAA/CIMSS ProbSevere and KTLX 0.5 degree reflectivity valid at 1828 UTC 07 May 2015.

Figure 2.  NOAA/CIMSS ProbSevere and KTLX 0.5 degree reflectivity valid at 1832 UTC 07 May 2015.

Seeing that rapid increase was noticeable to both of us as were were typing up other blog posts and prompted us the interrogate the radar data. While the core looked small initially, the idea shown by the ProbSevere was right on.

Finally, with the location of the cell being right under the clouds from upstream convection (see image below), it led to N/A values on the satellite data. 

-SRF

Initial Analysis localized as OUN

Water vapor imagery early this afternoon is showing a deep closed upper low over CA with broad southwesterly flow across the southern plains with a weak disturbance moving through the flow which seems to be enhancing/sustaining convection moving through the western portion of the forecast area. Visible imagery shows well the ongoing convection as well as a boundary situated across KS southwestward into the OK/TX panhandle.

A quick look at the GOES-R CAPE showed values generally in the 1000-1500 J/kg except along the Red River where lower values were shown below 1000 J/kg.  There definitely appears to be some influence of cirrus and mid level clouds associated with the ongoing convection affecting the output.

Comparing the GOES-R CAPE to the SPC mesoanalysis mixed layer CAPE it appears the values are too low, especially in the Red River area where values close to 2000 J/kg were noted.

The continued moist southeasterly flow into the region today south of the ongoing convection should interact with outflow boundary from ongoing convection and initiate new convection across the southern portion of the forecast area.

Jack Bauer

Rapid Increase in ProbSevere

While we were writing up a previous blog post, we noticed rapid increase in the ProbSevere for a storm to the west of the OUN radar. The ProbSevere quickly ramped up from 9% to 94% as seen in the two images below.

Seeing that rapid increase was noticeable to both of us as were were typing up other blog posts and prompted us the interrogate the radar data. While the core looked small initially, the idea shown by the ProbSevere was right on.

Finally, with the location of the cell being right under the clouds from upstream convection (see image below), it led to N/A values on the satellite data.

20150507_1830Z_Sat_Radar-SRF