Tuesday, May 12, 2015

Daily Summary: Week 2, Day 2 (May 12, 2015)

Today wins the distance award as we split up across the country to diagnose convective initiation in various environmental regimes. In the morning, forecasters started in Sterling, VA (LWX), Tampa Bay (TBW), and Pocatello, ID (PIH). Our LWX team used a suite of the GOES-R LAP and CI products to diagnose the near-storm environment and convective potential along/ahead of the cold front across the Appalachian mountains today. With minimal convection in the late afternoon hours, they were moved to Denver/Boulder, CO (BOU) to evaluation the pGLM color curves of some very marginal convection . Out PIH team also spent a good part of the day diagnosing CI and evaluating ENTLN tools and trends as a region of instability on the westward side of the upper-level ridge moved into the area. Our TBW team stayed quite busy with sub-severe seabreeze thunderstorms occurring across most of central FL throughout the day. ProbSevere and the suite of GOES-R products were used actively to target and prioritize regions with promising CI potential throughout the day.

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

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

GOES-R LAP
- First thing I pulled up yesterday, for cape product, it had a section of decent cape in Wilmington area, where a few storms did develop, did a good job of picking up on where convection did happen.
- It was very splotchy. Not realistic.. If it was more smooth, it would be more useful, it makes you question the quality with 2000+ j in area of 500.
- PW will be good for flash flooding situations.
- Closest sounding to my CWA today was pretty far away, so it was useful.
- Did 4 panel for comparisons, RAP was the big outlier. Matched closely with GFS and LAPS
- I think this will be helpful for winter weather

NUCAPS
- Idaho – see blog post. Adjusted using mesowest sfc conditions. I selected one of few points that were in the clear air. The sounding indicated modest instability in the area. One cell later strengthened as it moved through the area. This will be great for WR where observations are more scarce.

GOES-R CI
- Good day to test CI product, we had cu up along the whole mid atlantic coast, it was picking up many clouds, 10-30% mostly. I think the 15 min res hurts, as we’d get 10-30% coincident with a 35+ dBz echo. It was continuing probs even when we had returns on radar
- Not useful when looking at individual cu and trends, but more useful to look at area probs and trends
- This is a good tool to get an idea to where sea breeze t storms would initiate. It is often difficult to figure out where storms will initiate along a sea breeze. From a trend standpoint, it gives you an idea of where to look. It is helpful to give me general location of where convection will develop.
- May be helpful for Monsoon storms; will they set up over the east or the west of ridge tops?
- Would be helpful to have some sort of mask that tells you that the storm initiated, giving you an idea of why the probs disappeared.

ProbSevere
- In Idaho we had to change the color scheme, because the greys were really hard to see with mostly low probs in play.
- There were situations when the contour overlay blocked out the radar
- It did key in on the strongest storms in Idaho, which was ~36%. Perhaps a separate color scheme for the west would be necessary.
- Was great for situational awareness, providing a broad view of the whole cwa. If you have many storms, it helps me to key in on which storms I want to investigate first
- I used it quite a bit in Florida. It did a good job of showing storms that produced weather. I issued 5 or 6 SVS’s based on prob severe, instead of all tilts. It really did a good job of honing in on the relevant storms without looking at all the base data. 60 dbz at -20 when probsevere already up to 90
- Every storm that did damage had a probsevere around 90, around same time you would see a warning (Tampa). Looking at base data, would not have issued any earlier, probably less lead time with more marginal cases.
- Unrealistically high mesh in FL
- While at the lower probs, the satellite fields helped me to identify which storms were growing fast. The increasing trends, even low, told me to keep monitoring. The descriptors were helpful, especially the sat growth, qualitative ones.
- Once probs got high, it did a great job of discriminating. We had a couple high probs that produced severe.
- Use the time tendency trend in probs as guidance to perhaps issue, or prepare warning.
- Trend perhaps is just as or more important than absolute values

Lightning jump

PGLM
- I was seeing more LMA than eni lightning (very low levels of lightning), and seeing LMA first – out west in boulder

GOES-R Convective Initiation Nighttime Transition

The GOES-R Convective Initiation seems to have issues at nighttime. This even occurs with the use of infrared satellite data in the algorithm.

Here are a couple images from the transition from day to night:




From these images, it seems the convective initiation algorithm is probably more useful during the day. The coarser resolution of the infrared satellite data could possibly explain this problem.

- During the day, the CI algorithm utilizes information from both IR and visible satellite imagery. At night, the algorithm has no visible satellite data and must rely solely on the coarser IR imagery. This explains some of the changes in probabilities you might see in the transition from day to night - BL

Mahale

LMA Color Tables

This wasn’t the best day to look at LMA color tables because there weren’t many strong storms. I have used Flash Density in Huntsville  on strong storms so I am familiar with the default colormap. Below is an example of a weak storm with the LMA 1 vs LMA Default color table.

So here is LMA 1. One positive about this is because the scale between 0-10 has a lot of colors so those levels stand out quite a bit. This is useful in pulse storms and weaker storms. Above that you can see it is more graduated.



 Here is the same cell but with the default. You can see under 10 or so, it just looks white and doesn’t show any pinpoint of the updraft or much variation under 10.


So I think that LMA 1 is good for weak storms but then LMA default for stronger storms!

 

Lauren (and Regina)

First DTA over PIH

Finally got a cell to exceed 25 flashes/min (only lasted for a minute or two), which triggered a DTA (since there is a lower threshold in the west).


At the same time, ProbSevere was near the highest we had seen all day at ~36% with MESH of 0.83 in.



The OT algorithm also began to indicate a possible OT with the storm at roughly the same time.



Hard to say which of these products was the first to key in on the intensification as they all started showing signs at approximately the same time. However, the lightning time series does show a rapid increase just prior to 2245 (lightning jump?).



Issued a SPS based on the experimental products and base radar data.

Ertel

Snow Contamination on CI

Just thought I would show everyone what it looked like to see a snow contamination error on CI…



Lauren

LMA detection before ENTLN?

Switching to Colorado, there was a storm that the LMA picked up on while there were no ENTLN data and a scan later only a few…

This is from 2303Z where the LMA was picking up on something.



Then at 2308, the ENTLN data is finally picking up on “something.” It is really hard to see in this degraded image but there is one negative CG strike and then 3 “pulses.”



I usually see “some” in cloud lightning before I see any LMA detection so this was interesting to see especially with such a weak cell.

- But LMA's detect Total Lightning, which includes in-cloud lightning... - BL

Lauren

More CI Detection

Here is another set of images showing some shower development (there is no lightning).

2130Z CI product



2137Z Radar- not much to see, maybe “a” shower.



2145Z CI product- the highest probabilities have moved east



2147Z Radar.



2200 CI has increased in some areas. The yellow is over 60% while the green is 50%



2156Z radar is showing some shower developing



2215Z CI…still showing moderately high probabilities



2216 radar shows more showers




What is interesting about this example is that even when showers started showing up, even some that are near or above 35dbz, the CI is still picking up CI’s on it. This is likely because the CI product actually stops at a certain degree of glaciation which is roughly equal to 35 dbz. So CI doesn’t “know” about the radar and the dbz levels. It just shows that even though CI may still be showing probabilities, there could still be something on radar.



Lauren

Severe hail in NC

A storm in North Carolina took a while to get its act together, but then took off in a hurry! Going from 20% to 53% to 89% ProbSevere values. Strong satellite growth brought the probability to 20%, while the MESH was still very low (0.38"). However, this was certainly the highest probability of any shower/storm up to that point. A forecaster in Sterling, VA noted that she liked the trend upwards in ProbSevere. An official NWS warning was issued at 20:54Z, and 1" hail was recorded at 21:12Z (this was outside our forecasters' CWA). This is a case where the forecaster should be aware that there are no 'magical' thresholds in probability, but the strong satellite trends could still highlight severe potential (a 1 in 5 chance in this case, at 20:48) despite weak MESH.

ProbSevere, GOES-East IR window BT, and composite reflectivity (20:48 UTC)

ProbSevere, GOES-East IR window BT, and composite reflectivity (20:50 UTC)

ProbSevere, GOES-East IR window BT, and composite reflectivity (20:52 UTC)

-Cintineo

NUCAPS Sample

Tested NUCAPS sounding for southeast Idaho.  The point used is circled.  The visible image was close to the sample time, but there may have been some cloud contamination.  Adjusted the surface observations upward based on the nearest MesoWest surface observation.  It was difficult because a lot of the lower points needed to be adjusted to keep it realistic.  The instability seems a little high, but it could be localized.  Will see how the thunderstorms in the area develop over the next few hours.


Update: This thunderstorm moved over the sampled area about two hours later.  It peaked at about 55-60 dBZ Composite Refl and had a NOAA/CIMSS Prob Severe Model reading of 24% at that time.  The 5-minute ENTLN Lightning 3 km grid peaked at 11 with this storm.  There was no ground truth to verify what happened there.  But will note the complex mountain terrain in that area and how difficult thunderstorm modeling and forecasting is in Western Region (with local effects, etc.).  This was a decent test.




-Champion

Damage Reported in Cape Coral (Sig #3 using ProbSevere)

ProbSevere prompted a third significant weather advisory. In this example, the probability increased from 13% to 74% in 10 minutes:







The base data confirmed strong winds and small hail would be possible with this storm.

Cape Coral Police reported several trees and power lines down in Cape Coral and South Cape Coral areas. They estimated the damage occurred between 3:40 PM and 3:55 PM.

This is another example where the trends in ProbSevere can be extremely useful to warning forecasters to find storms that need further integration (“all-tilts without looking at all-tilts”).

Mahale