Daily Summary: Week 2, Day 1 (May 11, 2015)

New week, new participants, and a very different convective setup compared to last week. Today our forecasters were exposed to all that the EWP has to offer while operating in Wilmington, OH (ILN), Louisville, KY (LMK) and Houston, TX (HGX) ahead of a cold front that spanned from southern Michigan into central Texas. Forecasters quickly spun up on utilizing ProbSevere as a quick diagnostic tool to prioritize which storms to interrogate first. Forecasters were able to establish their first impressions, build procedures, and ask questions on the entire suite of products thrown at them. The established large squall line that dominated most of the country, while not helpful from the CI perspective, gave forecasters the ability to put the various diagnostic tools to work and compare outputs between mature and developing convection.

Overall, another "Day 1" in the books, here's hoping for some interesting cases this week (fingers crossed and prayers needed).

-Darrel Kingfield, CIMMS/NSSL Research Associate & Week 2 EWP Coordinator

Feedback from Daily Debrief
- Bill Line, SPC/HWT Satellite Liaison

GOES-R LAP
- PW fields compared well with observed soundings. This gave me a good feel for how the environment has changed since the morning soundings.
- It was helpful to see the three PW layers

NUCAPS
- I modified a profile near Houston, and CAPE was too high. There was a lot of cloud cover in the area, so it is hard to know if the profile was reliable.
- You can't just modify the surface values, you must modify the whole mixed layer, otherqise you get unrealistic lapse rates.
- I modified a sounding near Wilmington, and it matched well with the special 18z sounding (see blog post).
- There are no RAOB soundings in Huntsville, so I would like to use these more at home.
- I do not know if some of our forecasters (in Huntsville) will be comfortable with making the low-level modifications.
- Automating the modifications would be great, including the low-level mixing

GOES-R CI
- It seemed a little unsteady at times, with probabilities jumping up and down with no clear trends.
- We saw a few good cases, we had a lot of cu so it was picking up on a lot of the lower probabiltiies (blues). Ahead of the line, it hit on a storm that developed fast and later formed an OT
- The algorithm certainly suffers from the 15 min updates, as it is unable to "see" trends that are occurring between scans.

ProbSevere
- I like it, it gives an all tilts view without the all tilts, integrating the whole radar volume.
- Best use was to but it on the .5 degree elevation, watch trends in probsevere, and as the probabilities in a storm amp up, monitor that storm closer.
- When you have 30+ storms in your CWA, probsevere will highlight which storm you need to actually watch closer. Great situational awareness tool.
- The trends in the probabilities are easy to follow
- The trends seem real, with storms strengthening after increases in probabilities.
- I like the mouse-over in probsevere, and being able to see trends in the environmental fields as well.
- What else would you like to see in the readout?
--- I would like to see differentiation between hail and wind threats
--- Height of velocity or reflectivity core
--- Mid level convergence signatures, radial velocities, downdraft speed, descending reflectivity core. Basically, integrated volumetric calculations.

Lightning Jump
- It helps to visualize where the lightning is amping up, keying me in on the strengthening storms
- I stuck to the smallest scale (200 km^2)
- It did line up with ProbSevere, with increases correlating between the two.

PGLM
- FED is really cool stuff,. Being able to see where clusters of flashes are occurring is helpful. It made it easy to see strengthening updrafts.

ProbSevere analysis near New Orleans

The first day of the EWP has been pretty intense but very interesting. I focused on a couple of storms in Texas: the first just west of Houston died pretty quickly but the second, north-west of New Orleans, proved more interesting. One of my favourite products so far is ProbSevere but I’ve found that it can be inconsistent with the polygons and probabilities.

At 21:00Z (below) there’s a 55% probability polygon highlighted in a high 0.5 Reflectivity area. To the northeast of this is a 5% polygon shown.

Prob Severe Usability

Broadcast meteorologists operate under a number of constraints in developing severe weather situations. As a broadcast met forecasts in real time, products that display easy to understand forecast information are more likely to be used and incorporated into breaking severe weather coverage. Prob Severe as CI Probability are elements that can be easily incorporated to on-air analysis, as they guide the broadcast meteorologist and the conversation he or she has with the viewing public.

UFFSU

ProbSvr First Look

One of my favorite things today about the ProbSvr is the ability to make it a “quick look.” This image is after marginally severe weather a little earlier, so environmental wise in general, it isn’t favorable. But what if it was favorable and I was just glancing, waiting for things to pop? Like the example below, I can quick glance at this mess and see that all the ProbSvr is under 10% so I shouldn’t be worried for severe weather without having to look through various levels of radar data.

Lauren

Observed Radiosonde Data/NUCAPS Comparison

A special 18Z radiosonde launch was done at Wilmington, Ohio. The special launch allows for direct comparison to a dervived NUCAPS sounding.

Here is the observed radiosonde data:

Here is a NUCAPS sounding from nearby:

Note that the NUCAPS sounding is not representative, especially near the surface. The surface temperature is 77F and the dew point is 55F on the derived sounding. A nearby METAR close by the NUCAPS sounding was 85F/61F.

However, if the boundary layer temperature and dew point profile is modified using nearby METAR observations (85/61), the SBCAPE is more representative to the observed sounding (1761 vs. 1688 J/kg):

Therefore, it is critical to look at the near-surface temperature and dew point profile when using NUCAPS derived soundings.

Mahale

Shenandoah storm

A storm southwest of Shenandoah National Park (Viginia) has developed outside of the SPC's Marginal Risk polygon (Figure 1).

Figure 1: 20150511 2000Z outlook

Despite weakish MUCAPE and effective shear, and the satellite growth rates of ProbSevere being several hours old, and thus expired (see Figure 2), the probability of severe was still 82% at 20:34Z, 10 min before NWS in Washington D.C./Baltimore, MD issued a severe thunderstorm warning. The MESH at this point was 1.62". Multiple forecasters have mentioned today how ProbSevere can help focus their attention to developing "pulse" convection, or areas where severe weather may not be initially expected. This is the benefit of a "fused" product: where one observation platform isn't signaling severe potential, often another one can. Large trees were reported down from this storm at 20:48Z, 4 min after the NWS warning.

ProbSevere contour with VIS and radar reflectivity, near Shenandoah National Park in Virginia.

John Cintineo
UW-CIMSS

ProbSevere vs ENTLN Time Series

Here is a good example showing the similarities in trends of ProbSevere, ENTLN total lightning data, and radar data. At 1908 UTC, lightning flash rates neared a peak and ProbSevere was ~50%. At  1919 UTC, the lightning flash rates had dropped significantly as did ProbSevere. By 1925 UTC ProbSevere quickly increased to ~75% at the same time that lightning flash rates were rapidly increasing. Finally, by 1945 UTC, ProbSevere decreased to 9% and lightning flash rates plummeted. Soon after this the storm dissipated. These types of products could help forecasters “hold onto” or “let go” of a warning sooner.

Ertel

Day 1-Svr Prob Example

Hello everyone! This is day one of Week 2 of the Experimental Warning Program and we are getting situated and learning the products and making procedures. I personally was excited to try out the ProbSvr model as I think it could help with storms that pulse and that become severe or near severe within a scan or two- this is extremely useful in the northeast and even in squall line situations. While looking at it today in LMK’s area, we got a report of power lines down and I captured some of the ProbSvr data and I found it very interesting. I’ll step through the times.

1818z- This is as the northern storm was strengthening and the ProbSvr is hinting about 24%.

Next, at 1823Z, the ProbSvr bumped up to 61%

At 1825Z, the next scan, the 61% is still there (probably the same run).

At 1829, the ProbSvr has decreased once again.

Although I didn’t put velocity here, it isn’t overly impressive. Important to note is the rear inflow notch that looks like it is trying to develop.

At 1830z, the office received a report of power lines down with this northern storm.

In the training on ProbSvr, the examples were giving huge lead times when the model went above 50% but this isn’t always the case. I found this case interesting because it only gave about a 7 minute “lead time” before the first event. It looks like the storm surged/pulsed right at that time but noticed how the model surged back down before the event. How does a forecaster handle this? When it surges up and goes right back down? Would it change my warning process if it decreased next scan? It is still too early to say what the implication is but I will definitely be keeping an eye on these probabilities as I go forward!



Lauren

Best Practices for the use of the NOAA/CIMSS ProbSevere Model

The NOAA/CIMSS ProbSevere Model is an excellent tool for situational awareness–especially in situations with widespread convection. The tool essentially allows for a quick all-tilts view without looking at all the radar tilts because it integrates data from the entire radar volume (and assimilates other non-radar data as well). For example, overlaying the tool on the lowest radar elevation (e.g., 0.5 degrees), you can monitor trends and areas of interest by the rate of change of the probability of severe. Rapidly increasing probability of severe gives the warning forecaster insight that the storm is rapidly intensifying aloft. At this point, a forecaster can do further interrogation on the storm of interest.

Here’s an example where there is widespread convection in southeast Texas:

Note the probability of severe rapidly increases (10% to 86% in 8 minutes on one storm) across the western storms. This immediately signals to the warning forecaster that these storms need further interrogation and are the storms of interest.

Mahale

Week 2 Day 1 - Mid-south convection

Storms are building along and ahead of a cold front from Michigan to Georgia, where there has been some clearing of clouds for adequate diurnal heating. Figure 1 shows the NOAA/CIMSS ProbSevere model output from a storm in central Tennessee, just south of one of our operating areas today. The moderate to strong satellite growth rates, ample MUCAPE, modest effective shear, and 1" MESH contributed to an elevated probability of severe, of 71%. The NWS issued the first severe thunderstorm warning at 18:56 UTC, 8 min later (Figure 2). The highest probability achieved was 83%. No reports have been recorded yet.

Figure 1: ProbSevere contours overlaid visible satellite and radar reflectivity.

Figure 2: NWS warning with ProbSevere and radar and visible satellite imagery.

-Cintineo