Week 1 complete!

Week 1 (of 4) of the Hazardous Weather Testbed 2016 GOES-R/JPSS Spring Experiment is now complete! Although there were no widespread severe weather outbreaks this week, the marginal events provided just enough to keep our participants busy. The week 1 participants provided plenty of useful feedback on all of the products demonstrated, much of which can be found here on the blog. Below are a few photos from the week.

Bill Line, SPC/HWT Satellite Liaison

Week 1 HWT GOES-R/JPSS Participants - Lef to right: Randy Bowers (WFO AMA), Chris Gitro (WFO EAX), Mike Jurewicz (WFO BGM), Bill Line (OU/CIMMS and NOAA/SPC), Antonia Gambacorta (JPSS), Dave Aguilera (KCNC-TV Denver), Kristin Calhoun (OU/CIMMS and NSSL). Justin Sieglaff (UW/CIMSS) was also in attendance.

Hard at work inside the HWT

Our 4 forecaster participants interrogate GOES-14 1-min visible imagery in the HWT.

On Tuesday, Chris Gitro (pictured) and Mike Jurewicz both presented in the SPC conference room to SPC and WFO OUN staff on their use of GOES Sounder Water Vapor imagery and Layer PW products in operations.

On Thursday, Mike Jurewicz (pictured) and Chris Gitro presented some of their dual-pol research in the WDTD conference room.

Week 1 (18-22 April 2016) Summary and Feedback

For the final day of the week, our groups participated in the Lincoln, St. Louis, and San Antonio CWA's. Below are final comments from participants recorded during the end-of-the-week debrief.

LAP
- I would like to see theta-e and theta-e difference computed from this.
- I would like the LPW layers to different (sfc-850, 850-700, 700-500, 500-300). 500-300 is valuable for diagnosing heavy rainfall, extending to 700 washes the signal out. You also might see ll dry layers better in the 850-700 layer.
- There needs to be a strong push to make LPW baseline with GOES-R
- Patterns in the fields matched up with what I expected.
- It was cool to see moisture evolution tracing the terrain
- Convection typically initiated along the gradients.
- We thought the fields looked really good.

CI
- All like the idea of having a regular CI and severe CI
- In Illinois yesterday, marginal fairly widespread mostly sub-severe convection. Days like this is where Severe CI would be helpful in telling us that developing convection would remain non-severe.
- CI is good for a quick glance, then I dive deeper into other tools.
- sub-30% just cluttered up screen, so I would make this transparent.
- Sometimes the display was too overwhelming.

Probsevere
- I didn't like the color curve at first, but by the end of the week I really liked it.
- Dual-Pol fields might help with Tor probabilities, theta-e difference for winds
- I envision a 4-panel display with ProbSevere, ProbWind, ProbHail, ProbTor
- A Hydo ProbSevere would be helpful.
- It would have to be alittle more dynamic in terms of predictors used in varying regimes. It didn't seem to perform well in High CAPE, low shear situations. Perhaps in such cases, less emphasis should be put on MESH.

1-min imagery
- I was fine with the appearance of the parallax-corrected imagery

Satellite-derived winds
- Yesterday the winds confirmed a mid-level jet coming around the base of the low in the 600-700 mb layer.
- These could be helpful in depicting the nocturnal LLJ
- These could be helpful for hydro situations, especially with terrain.
- I like the idea of deriving LL shear from the surface obs to the visible winds. This could be helpful for TAFS also.
- Models generally don't do very well with winds in the low-levels, making these important.

Lightning Jump
- I thought it had predictive skill for severe convection
- I'd like to see a transparent or contour display
- I think I would mostly look at the 5-min max. It would be nice if this told me the exact time of the jump.
- I'd like to see the addition of a trend graph.
- Some sort of table or a separate app to monitor products, including ProbSevere, in real-time would be nice.

GLM
- I recommend contouring flash density, as opposed to the pixel look, for overlaying purposes.
- A colortable simialr to ProbSevere would be a good option. Brighter display with increasing density.
- I like to overlay lightning data on radar or satellite.
- It will be helpful to have condensed, quick reference material available for using lightning data while on shift.
- Climo that will come out of GLM will be helpful.
- Some blended total lightning product would b helpful
- Offices may need a lightning FP given all of the different lightning data. This is a training issue.
- NWS forecasters are not used to incorporating lightning into their forecast products, they are behind the curve. Part of the problem is training, part is people needing to get out of their comfort zones.

NUCAPS
- Modifications to the sfc/low-levels are a necessity. We need to see the CAP.
- In it's current state, I think it is a tough sell. But if we are able to incorporate the automatic adjustments, it'll be easier to show others the benefits.
- I would like to see weighting functions/sensitivity functions for the information used in NUCAPS>

Lightning Probabiltiy Difference Scale1 vs Scale 0

Note: The CG Lightning Probability product is not a GOES-R or JPSS funded project.

Not many lightning strikes going on so far today. Lightning probablility is showing strikes that are occuring. The difference between scale0 and scale1 seems to support the idea that scale 0 might give more specific information over scale one on which storms have the best lightning probabilty. Scale one might be useful for overall early coverage and scale 0 might be a better indicator as more storms develop. 
 -thunder

Lightning Probability scale 1

                                           Lightning Probability scale 0

NUCAPS Planviews

Overall, the NUCAPS planview maps would be beneficial in the forecasting environment as added temperature data would be available inbetween standard upper-air launches.  This could serve as a good proxy to help judge the strength of a capping inversion, while also possibly serving as an additional information source during winter wx events.  Is there a way to interpolate between QC-flagged data?

ILX Severe Thunderstorm Warning

Thus far there have been numerous thunderstorms over the ILX CWA, with a lot of storms likely producing pea sized hail.  The ProbSevere model began to highlight a storm over the far northeastern CWA, reaching probabilities much higher than all surrounding storms.  Figure 1 shows 0.5 degree ILX reflectivity and NOAA/CIMSS ProbSevere contours.  The ProbSevere read-out shows the storm is in a moderately sheared environment (~35 knots of effective bulk shear) and the probability increased as the MRMS MESH increased to over 1.00".  The ProbSevere trends along with all radar tilts allowed the forecast team to issue an experimental severe thunderstorm warning at 2159 UTC.

Figure 1.  0.5 degree ILX reflectivity, NOAA/CIMSS ProbSevere contours, and experimental severe thunderstorm warning polygons.

The forecasters noted there have been many storms today in the ILX CWA with appropiately low ProbSevere values, but the ProbSevere model did a good job highlighting this isolated storm that became much more intense and required a warning.

UPDATE:  This storm produced 1.00" hail at 2215 UTC in Gibson City, IL.

-Sieglaff

Merits of Prob Severe

J-Money:

For much of the afternoon, so far, scattered to numerous storms in central/southern IL have been non-severe in nature, with occasional small hail.  Prob Severe values have been correspondingly low, with meager flash rates, and poor growth rates.

However, just before 22z, a storm intensified in the far northeast part of the ILX CWA.  Prob Severe values rose quickly into the 70s and 80s, as flash rates and MESH values increased.  A warning was issued, based on the environment of cold air aloft, and seemingly more favorable storm-scale parameters for hail, as shown by Prob Severe.

A loop of refectivity at the -20C level, with Prob Severe overlayed, is shown here.  It well correlates with the above descriptions.

 

NUCAPS vs. Observed and Forecast Soundings

On the large scale Thursday afternoon, northern Oklahoma was on the backside of a strong upper low pressure system which was moving through the upper Mississippi Valley.  On the backside, low, fair wx CU was readily apparent across southern Kansas and northern Oklahoma.  Further aloft, water vapor imagery showed an influx of dry air sinking south across the Central Plains through the early afternoon hours.  This dry air would eventually move into northern Oklahoma and was sampled by a 17z sounding from Lamont, OK.  The NUCAPS pass at 19z also sampled this dry air very well, as highlighted below.  Additionally, the NUCAPS sounding also accurately sensed an increase in moisture near the 850 hPa level which per the visiable satellite image above, was very likely the flat fair wx CU field seen on visible satellite imagery above.

Recommendation:

Through the course of the first week of the HWT, we've learned that there are plans for automated correction to the NUCAPS soundings based on surface observations and rapid-refresh analysis.  If I were to make a recommendation in addition to the boundary layer correction, I think the best area to target would be the 850-500 hPa layer as this is the area where capping inversions are commonplace.  Knowledge on the strength of a capping inversion is crucial in the severe weather forecasting environment and NUCAPS soundings can provide added value in the near-term convective forecasting environement.  

19z NUCAPS Sounding

20z Rapid-Refresh Sounding

17z Lamont, OK Sounding

NUCAPS Sounding Comparison

The 17z sounding at Lamont gave us an opportunity to test the NUCAPS soundings again. This time the accuracy was more acceptable. The strengths of NUCAPS were apparent. The first image below is the Lamont sounding valid at 17z. Below that image is a water vapor loop showing an area of dry mid-upper air advecting into northern Oklahoma by the time of the 19z NUCAPS sounding. For reference Point A was placed over Lamont.

 

Below is the NUCAPS sounding. You can see that it captures the mid-upper dry layer well.

 

You can see in the two images below that the RUC shows the trend but may not be pronounced enough with the mid-upper dry layer, and this is the one clear strength of NUCAPS. Perhaps there's an opportunity to augment the mid and upper moisture in the models with NUCAPS data. However, there still seem to be errors that have operational implications for the thermal and moisture profiles lower in the atmosphere.
 
 

-dryadiabat

NUCAPS comparison for IL

J-Money:

A NUCAPS sounding was sampled over southern IL, east of St. Louis, and modified a bit for the surface-700 mb environment.

After modification, as compared to the SPC meso-analysis, the SBCAPE was a bit lower in NUCAPS (750 vs. 1000), with the MLCAPE also a bit lower (200 vs. 500-600).  See attached.

Severe warning lead-time in Brownsville WFO

While thunderstorms are remaining well below severe thresholds over the Middle Missippi Valley, a strong to severe thunderstorm initatied in the Brownsville WFO along residual outflow Thursday afternoon.  Figure 1 shows ProbSevere contours along with GOES visible imagery and MRMS composite reflectivity from the ProbSevere website zoomed on the storm in the Brownsville CWA.  The storm was in a favorable environment and exhibited strong satellite growth rates and glaciation rates, with a beginning probability of 28% at 1936 UTC.  The probability quickly increased (exceeding 80% at 1952 UTC) as lightning flash rate and MRMS MESH increased through 2014 UTC, when the first severe thunderstorm warning was issued.  This illustrates how ProbSevere can extend lead-time and increase confidence.

-Sieglaff

1-minute imagery + CI + satellite winds giving a heads-up for near-future storms?

J-Money:

Relatively clear areas over central/southern IL so far today, may see increasing convective development the next 1-2 hours.

1-minute visible pictures and regular CI images have highlighted an agitated cumulus field near the MO/IL border, with radar showing scattered storms.  At the same time, satellite derived winds confirm strengthening SSW flow coming into IL, around the cold-core vortex over southeast IA/northeast MO.  This matches well with SPC meso-analysis data.  See attached image.

Convective Instability increasing as seen on Layered LAP PW data

The image below shows a well-defined dry layer advancing east from the lower Missouri Valley into the Mid-Mississippi Valley.  In particular, the 700-300 hPa precipitable water image (upper right panel) shows very dry air working east over low-level air that is characterized by higher moisture.  This signals an increase in convective instability across the Illinois and Mid-Mississippi River Valleys.

Setting up in Lincoln, IL and St. Louis, MO

In a change from recent days, we've moved north today and will be focused  initially on Missouri and Illinois.  The main hazard today will be marginally severe hail associated thunderstorms ahead of a once cut-off low that is now ejecting east-northeastward toward the southren Great Lakes region.

-Sieglaff

Daily Summary: Week 1, Day 3 (April 20, 2016)

We continued to work in Texas today, with groups operating in Lake Charles, Amarillo, San Antonio, San Angelo, and Lubbock.

LAP
- I thought the LAP fields were pretty useful. They were good in showing the moisture gradient over central Texas, showing moisture return back into the panhandle ahead of a synoptic-scale feature. I could see moisture moving through the topography, which was neat.
- I glanced at CAPE, the spatial evolution seemed good, but the absolute values, not so much.

CI
- A lot of False Alarms again yesterday.
- I did not see Severe CI hit on much. It did hit one storm hard in Midland.

ProbSevere
- One storm produced ping pong ball sized hail, and ProbSevere hit on it. Probs hung out in the 60-70 range, then took off so I warned.
- I am learning how to use ProbSevere more intelligently now. I think it is important to pay attention to the ingredients, and why the probabilities are trending up and down. Not focusing just on probs, but why.
- I liked to overlay it on reflectivity at -20C level
- A storm in Amarillo was driven more by lightning, while a storm in Lubbock was driven more by radar. I tis nice to see how different fields can drive the probabilities.

1-min imagery
- It was helpful to compare the 1-min imagery with 15-min data to see what additional information I am getting
- I use it just as frequently to rule out imminent storm development.

Satellite-derived winds
- I see this being helpful in diagnosing LL shear, and trends in shear, from the surface to cloud base.
- Forecasters thought that derived analysis fields such as wind speed and divergence would be useful.

PGLM
- I liked overlaying 1-min satellite imagery, flash density, and lightning jump. However, the overlay did block important details in the satellite imagery. It might be better to do this as a 4-panel.

Lightning Jump
- Last 2 days, I got on the order of 5-10 min lead-time storm intensification
- It does often fluctuate between low and high jumps, but the high jumps do attract my attention.
- I trend graph of recent LJ's would be useful

NUCAPS
- I had to do quite a bit of manipulation at the low-levels, as the whole BL was too cool.
- All participants thought an autmated blend with RAP and sfc would be ideal
- If we got to the point where the manual manipulation is streamlined, and you use the strengths of bboth RAP and NUCAPS, that's a lot of soundings at your disposal. Right now, it is tough with the amount of effort that goes into the modifications.
- There is no inversion present in NUCAPS, which is a big negative.

ProbSvr gives good lead time

Looking at the use of prob severe in providing lead time. At 23:36z a report of ping pong ball size hail.

Analysis of prob severe and lightning jump 20 minutes before at 23:16z , shows the probSvr displaying 91% chance of severe, 46kts shear, MU cape 947, Mesh 1.07", Growth rate strong: glaciation: strong, flash rate 28/min and 1 signa jump.

20 minutes later at 23:36z:  90% chance of severe, 49 kts shear, Mu cape 1157, Mesh 1.93", Growth rate:strong, Glaciation: strong, 0 signa. ( however, jump did stay at 1 during 20 min duration.)

One report of ping pong ball size hail at that point.

Prob severe may be a good factor in issuing an early warning for a severe storm.

-thunder

Another Prob Severe Success East of Amarillo

This storm east of Amarillo produced quarter size hail at 6:25 PM, approximately 11 minutes before severe prob increased above 80%. There was a steady increase in values over the 30 minute before the severe hail report, from 20% at 6:02 PM, to 25% at 6:08 PM, to 62% at 6:12 PM. Correlated with the time of the severe hail report, severe prob values were 90%. Glaciation rates went from weak to strong ~20 minutes before the severe hail reports. Flash rates steadily increased leading up to the severe hail report but peaked ~30 minutes after the report. 

-dryadiabat

Interrogating 1-min Satellite Imagery in the HWT

This week's group of HWT participants have enjoyed interrogating the 1-min satellite imagery from GOES-14. We have already had many great discussions regarding the various phenomena observed.

The 1-min data has proven to be a valuable dataset in the experimental warning operations. Forecasters are specifcally highlightning features and processes that are not readily apparent in the currently avaiable operational imagery. Some examples include:

1) Gravity waves, especially subtle, often eminating from developing and ongoing convection,
2) better tracknig of cloud character and trends, especially contrasting cu that have become agitated,  
3) earlier detection of convective initiation and failed updraft attempts,
4) anticpating boundary interactions,
5) identifying and tracking overshooting tops, many of which occur on time scales less than 5 minutes,
6) the 1-min data has been discussed as having additional value when the local radar is in SAILS or meso-SAILS mode as it provides continued information about updraft strength.

Particpants have commented that they are seeing processes that they have never observed before in satellite imagery, and over time, will learn much from such imagery.

Ping pong ball hail in Lubbock CWA

Back in the Lubbock CWA, we were monitoring a storm for a long period of time southeast of Lubbock.  The ProbSevere values fluctuated over time, generally in the 50-70% range.  Suddenly at 2338 UTC, the ProbSevere value jumped from 65% at 2336 UTC to 88% in response to an increase of the MRMS MESH from 0.76" to 1.15".  Additionally, this storm had moderate vertical growth rates, strong glaciation rates, and was an environment of very high effective bulk shear (~53 knots).  Figure 1 shows the progression of ProbSevere values from 2336 UTC to 2345 UTC when ping pong ball sized hail was reported by an NWS employee, or about 8 minutes of lead-time from the 65 to 88% jump.

Figure 1. 0.5 degree KLBB reflectivity and NOAA/CIMSS ProbSevere contours valid 2336 - 2344 UTC 20 April 2016.

-Sieglaff

LBB Storm continues to move southeast

Storm very briefly pulsed up and dropped ping-pong sized hail east of Post, TX.  PROBSvr caught this very well as the MESH increased along with probabilities of severe increasing to 90%.

Eastern Amarillo CWA Storm

While we are watching storms in the Lubbock CWA, I kept an eye on a storm over the eastern portion of Amarillo CWA, near I-40.  The ProbSevere was 25% at 2308 UTC and jumped to 84% by 2314 UTC as MRMS MESH and total lightning flash rates jumped significantly (also the effective bulk shear was rather high in the mid/upper 40 knots).  The ProbSevere values continued to increase into the 90% range.  A 1.00" hail report was received at Interstate 40 at 2325 UTC and after the report was received, a severe thunderstorm warning issued.  The good lead-time for this storm can be attributed to all predictors of ProbSevere--strong satellite growth rates, strong satellite growth rates, increasing lightning in a highly sheared environment and MRMS MESH over 1.00".

Figure 1.  NOAA/CIMSS ProbSevere contours and MRMS composite reflectivity valid 2308 - 2332 UTC 20 April 2016.

-Sieglaff

Tracking Low Level Moisture Trends

Notice that the deeper mixing and developing/sharpening dryline in the southwest Texas Panhandle is evident in the lowest of the layer precipitable water. This has great value tracking low-level theta-e ridges, drylines, and other moisture discontinuities.

-dryadiabat

Experimental CG Lightning Probability

Note: The CG Lightning Probability product is not a GOES-R or JPSS funded project.

Below is an animation of MRMS radar 0.5 km reflectivity with the experimantal CG lightning probability and NLDN CG overlaid. Notice that the experimental CG lightning probability rises to 58% 14 minutes before the first CG strike detected by NLDN. Probablities increased to 67% 5 minutes before the strike, and 72% approximately 2 minutes before the strike. This has potential significant value for providing lead time to outdoor events. In addition, the National Weather Service in Amarillo recently began experimenting with Aviation Weather Warning for lightning for Rick Husband Airport, and this experimental product could potentially allow the office to provide a better service to aviation customers. Other offices provide lightning warnings to airports and likely could potentially benefit from this. I will be watching this product closely throughout the rest of this experiment to see how much lead time it normally provides and how frequent false alarms are. My first impression is very possitive.

-dryadiabat

Strom Southeast of LBB's RDA

Storm development has occurred southeast of the Lubbock, TX RDA in the Texas Panhandle.  While initially the all-tilts reflectivity looked promising, PROBSvr continued to indicate marginal severe potential.  Combine this with no significant lightning jump and low flash density on the PGLM array, the decision was made to hold off on a warning for now.

Moisture return (as indicated by Layered PW products) portends convective intensification in the TX Panhandle

J-Money:

As the attached images show, a narrow S-N plume of moisture is returning to the TX/OK Panhandles, coincident with the approach of a 700-500 mb short-wave.  As such, convection is intensifying.

                                                  1-minute visible loop

                                                 4-panel layered PW

                                          SPC meso-analysis at 850 mb

CI Prob leads convection

CI Prob signature jumped to 80 before convection followed up 15 minutes later on vis imagery and radar. -thunder

   CI Probablilty loop

  Radar Loop

Severe CI false alarm

J-Money:

On the upper right-hand panel (Severe CI) of the attached image, at about 2130z, a very high probability of severe development spiked (red color, likely at or above 80%).  This, otherwise, seemed like garden variety cumulus development, enhanced by the terrain.  The cumulus field in this vicinity subsequently dissipated/flattened, between 2130z and 22z.

One Minute Data in IR and WV

One minute data is not limited to visible satellite. Here's a comparison of one and 15 minute data for IR and WV.

Reverberating effect of a thunderstorm, as shown by 1-minute imagery

J-Money:

As this loop shows, the high-res visible imagery deptics a ripple effect of clearing/stabilization, in almost every direction from this strong-severe storm.

1 minute imagery

1-minute visible shows so much info. While we are looking at the cloud features. We see what may be a gravity wave pushing over ne. Wave clouds pushing againt northern flank of building T-storm and what looks like diffluent flow pushing into storm from the SE which could slow storm strengthening. Lots of features that would not be as prominent in normal imagery.
-thunder

1-minute imagery and near-term storm evolution

J-Money:

The attached loop shows the storm perhaps becoming anchored to an outflow boundary.  Better stream-wise vorticity potential?  Three cheers for 1-minute data!