Forecaster Thoughts – Ashlie Sears (2013 Week 3)

The 2013 EWP was held for three weeks in May, with 6 participants per week attending from Eastern, Central and Southern Regions. I had the opportunity to attend in the third and final week. Unfortunately, many issues arose during my week which affected the initial timeline/plan laid out for us to follow. The first of two EF5 tornadoes to hit the OKC area ravaged Moore, OK on the Monday. We also were limited on the tools to test Wed-Thurs with the loss of GOES-14.

The test bed was arranged to where there were two forecasters on a meso desk for the duration of the shift each day. They would provide hourly updates on the environment or determine if the area of concern needed to be shifted. There were then two teams of two who were given responsibility of different CWAs and providing warnings for their respective area.

Timeline for the Week

Sunday – Arrived in Oklahoma.

Monday – Introduction to the test bed, overview of the week, discussion of expectations for the week. The Moore tornado occurred at the beginning of this shift and with many people helping to run the test bed being affected by the storm (i.e. their kids in the path) along with technical issues we were dealing with, there was a lot of confusion to the first day, limiting to what we were able to accomplish. We were unable to use any of the test bed products in real time to analyze the tornado that occurred, though we were able to go back later in the evening and analyze the day.

Tuesday – Warning Operations over Fort Worth and Shreveport CWA early on, followed by Albany CWA in the evening.

Wednesday – Warning Operations over Buffalo and Binghamton CWA.

Thursday – Warning Operations over Amarillo, Lubbock, San Angelo and Midland CWA.

Friday – Week wrap up, with a 30 minute national call discussing things learned during the week.

Lessons Learned

With a shortened testing period this year, each week was asked to analyze all the tools to be tested, though focusing on two primary areas. We presented nationally for the “Tales of the Testbed” webinar on the last day covering these two specific topics, discussing best practices and how we found we could utilize in the tools in our own office operations. The focus for my week covered the Multi-Radar Multi-Sensor (MRMS) and the Hail Size Discrimination Algorithm (HSDA). In addition to the tools/experiments, we were asked to provide feedback on the usage of AWIPS 2 and the whole warning generation.

Mesoscale Analysis Tools

There were two things we were asked to utilize in determining the mesoscale situation. Unfortunately, these tools are restricted to their domain of Oklahoma and Northern Texas and that general vicinity. However, talking with the representative for the LAPS forecast, I was informed that they eventually would like to spread to the rest of the country, depending on the success they find in the southern Central Plains.

OUN-WRF and LAPS 1km and 3 km forecasts cannot be utilized up here in the Northeast. However, the LAPS 2.5 km analysis is available to be used at this time. I found during our warning situations that using the Theta-E values from LAPS was a best practice in figuring out where the convection was going to initiate and continue to form. The ability of AWIPS2 to allow multiple layers to be easily overlaid and the easy access with zooming in and out without constantly having to reload  the frame allowed the forecaster to analyze the situation at a much quicker pace as well as obtain a much clearer picture. I found it beneficial overlaying the analyzed 2.5 km CAPE values with the latest reflectivity images then comparing to how the LAPS 1 and 3 km forecasted these parameters over the next few hours. Overall, the LAPs had the better grasp of the type of storm development and the location though the timing was off by up to 1-2 hours. It also is run at 15 minute intervals, allowing a more up to date version of what is occurring. The OUN-WRF had a better handle of the timing of the development, but was off on the location.

I would really like to see the forecasted LAPS in our region of the country as it proved quite useful in determining where the most likely area of convection would be. In addition, in discussing other potential uses, the LAPS rep informed me they have found it useful in forecasting the rain/snow line, producing fairly accurate forecasts, during winter events which would also be very useful in the Northeast.

Convective Initiation Tools

Unfortunately I wasn’t able to test these tools as well as the other experiments. On Tuesday, we arrived after initiation had begun, so we were unable to do any pre-event analysis. On Wed, we lost these tools when the GOES-14 satellite was down and our warning domain was outside GOES-13’s track. Luckily Thursday we were able to analyze the CI tools for the storms that occurred over northern Texas, allowing some familiarity but not as much as I had hoped. One side note I would like to make, I have been analyzing these tools available via the web to see how each handle CI here in the Northeast. For the couple of convective events we have had in the past few weeks, I have found that the CI tool has given about 1 hour lead time for areas further inland, but is still having issues with capturing the potential for convective development along the coast. One issue these tools do have is cirrus contamination and this was noted in several areas we were analyzing during the test bed.

Warning Tools

The experiments that were the focus of the week were the MRMS and the HSDA. The incorporation of the MRMS and HSDA will be very beneficial in warning operations. These products combine data from multiple sources, doing a QC over the data and then producing a final result that depicts a clearer image of what is actually occurring, compared to the usage of just one radar/source.

One of the more interesting aspects of the MRMS was the ability to use something other than reflectivity and velocity to create your warning polygon. For the issuance of tornado warnings, using either the 30 minute or 60 minute tornado track allowed the forecaster to see the trend of rotation, the strength of it as well as the path it was taking. While we may have to have four panels or more up to see rotation throughout the vertical column, using the rotation track products allowed us to see general rotation strength in the lower levels as well as in the mid-levels, just by looking at two screens. The color coding of the product also gave us an indication of the severity of the situation and caught our attention easier than diagnosing an all-tilts product.  The same concept was used with the MESH tracks which allowed the forecaster the ability to see the track of the hail core producing severe hail as well as enabling the forecaster to just look at one product (or multiple depending on the time scale) versus having to have an all-tilts up, which takes space and time. One other aspect I found very beneficial with the MRMS was the ability to set up a four panel screen with reflectivity at the 0oC, -10oC and -20oC compared to the lowest level reflectivity scan. This allowed for a quicker diagnosis to determine if severe hail was being produced within the hail growth zones. All the while too, because these products are incorporating many different sources, it also allows the forecaster to not to have to trouble themselves on bringing up all the surrounding office radars in addition to their local radar. This could help in saving memory space within AWIPS, allowing for the program to run quicker as well as general diagnosing time for the forecaster.

The HSDA products also provided quite valuable in determining the hail potential of a storm. Once it is determine if an area of the storm is possibly producing hail, the use of the HSDA then informs you of the potential size of it. We had great success in verifying the large versus giant hail in the storms of the southern Central Plains. Unfortunately, we were able to use the product for the storms here in the Northeast due to a misunderstanding of the use of the product. But I would be curious to see how well the storms verify compared to the HSDA in storms up here as well.

AWIPS2

Overall, I had great success utilizing AWIPS2 into the warning operations during the test bed. The one main issue that occurred when I was producing warnings was that if a warning polygon happened to overlap into a different CWA when I was first constructing it, even after hitting the Warned/Hatched Area button, it still issued the full polygon that overlapped into the other CWA. This was presented to the IT rep during the test bed to hopefully be corrected. Otherwise, the functionality of AWIPS2 really did speed up the analysis as well as warning process.

Ashlie Sears
General Forecaster
NWS Upton NY (New York City)
2013 Week 3 Evaluator

Tags: None

EWP2013 Week 3 Summary: 20 – 24 May 2013

EWP2013 PROJECT OVERVIEW:

The National Oceanic and Atmospheric Administration (NOAA) Hazardous Weather Testbed (HWT) in Norman, Oklahoma, is a joint project of the National Weather Service (NWS) and the National Severe Storms Laboratory (NSSL).  The HWT provides a conceptual framework and a physical space to foster collaboration between research and operations to test and evaluate emerging technologies and science for NWS operations.  The Experimental Warning Program (EWP) at the HWT is hosting the 2013 Spring Program (EWP2013).  This is the sixth year for EWP activities in the testbed.  EWP2013 takes place across three weeks (Monday – Friday), from 6 May through 24 May.

EWP2013 is designed to test and evaluate new applications, techniques, and products to support Weather Forecast Office (WFO) severe convective weather warning operations.  There will be three primary projects geared toward WFO applications this spring, 1) evaluation of multiple CONUS GOES-R convective applications, including pseudo-geostationary lightning mapper products when operations are expected within the Lightning Mapping Array domains (OK/west-TX, AL, DC, FL), 2) evaluation of model performance and forecast utility of the OUN WRF when operations are expected in the Southern Plains, and 3) evaluation of model performance and forecast utility of the 1-km and 3-km WRF initialized with LAPS.

PARTICIPANTS:

Our participants included Eric Martello (WFO Fort Worth, TX), Ashlie Sears (WFO New York, NY), Jeremy Wesely (WFO Hastings, NE), Kris White (WFO Huntsville, AL), and Andrew Zimmerman (WFO Wakefield, VA).  The GOES-R program office, the NOAA Global Systems Divisions (GSD), and NWS WFO Huntsville’s Applications Integration Meteorologist (AIM) Program have generously provided travel stipends for our participants from NWS forecast offices nationwide.

Other visitors included Jason Otkin (Univ. Wisconsin), Steve Albers (NOAA/GSD), Chad Gravelle (NWS Training Center GOES-R Liaison), James McCormick (Air Force Weather Agency, Omaha, NE), Lori Schultz (University Alabama – Huntsville), and Jim Gurka (NOAA/NESDIS/GOES-R).

Greg Stumpf was the weekly coordinator.  Clark Payne (WDTB) was the “Tales from the Testbed” Webinar facilitator. Our support team also included Darrel Kingfield, Gabe Garfield, Travis Smith, Chris Karstens, Kristin Calhoun, Kiel Ortega, Karen Cooper, Aaron Anderson, and David Andra.

REAL-TIME EVENT OVERVIEW:

20 May:  Norman (OUN), Fort Worth (FWD):  Post-Moore Tornado supercell storms in southern OK and northern TX.  NOTE: The Moore tornado occurred too early in the shift and network outages cut off our live data feed.  However, the historical track products (e.g., Rotation Tracks, Tornado Debris Signature Tracks) were viewed after the event.

21 May:  Shreveport (SHV), Fort Worth (FWD), Albany (ALY), Binghamton (BGM):  Severe storms with hail in TX and LA, followed by marginally severe storms in upstate NY.

22 May:  Buffalo (BUF), Binghamton (BGM):  Supercells produced large hail in upstate NY.

23 May:  Amarillo (AMA), Lubbock (LUB, Midland (MAF):  Supercells with large hail and a few weak tornadoes in the TX Panhandle.

FEEDBACK ON EXPERIMENTAL PRODUCTS:

HSDA:

  • A logical progression of the HCA product, but useful than the current version.
  • Works well with MESH products, although seems to underestimate the hail size versus MESH.

MRMS:

  • Likes instantaneous MESH / height about -20C / -10C (lightning)
  • Tracks difficult to use for slow-moving cells; best for fast-moving storms.
  • A lot of the MRMS stuff makes derived stuff obsolete.
  • Track orientations give a lot of utility.
  • MRMS was considered by some forecasters to be the most useful new products for warning ops.
  • One forecaster thought it was outstanding and will be heavily used in the field.
  • Rotation Track product for the Moore tornado could be used to help determine operational warning thresholds.
  • The 2-minute temporal resolution is advantageous, potentially buying lead time.  This could potentially keep forecasters from switching back-and-forth from radar to radar and saving time from looking at all-tilts so much
  • Helps with the overall SA and the efficiency of the warning process.  With multiple deep cells in the area, products like the MESH, and -20 C Reflectivity can be very helpful for determining the cells on which to focus.
  • Was useful during a temporary radar outage, as data from adjacent radars was used to fill in the storm without having to analyze separate radar feeds.
  • Offered a quicker diagnostic overview of the storm than having to look at multiple height scans as we commonly do with the Donovan method.
  • Storms near or over a radar within a cone-of-silence had data which was effectively filled in with other radars.
  • Verified that MRMS products matched very closely the values one would get using manual data interpretation using all-tilts and sampling.

OUN WRF:

  • 0-1 km SRH fields received a lot of praise.
  • Recommend adding an updraft helicity track.

Variational LAPS:

  • The 15-minute temporal resolution of the product can be very useful for diagnosing locations of continued convection especially in rapidly developing convective situations.
  • Overdid the storms and created too much outflow.

GOES-R Simulated Satellite:

  • Helps to better visualize how the model is creating and evolving convection.
  • Recommend a product or procedure that will facilitate a comparison to real data.

GOES-R RGB Airmass:

  • Training on RGB airmass needed to be made more adequate.
  • Best when used with other environmental products to see how conducive the airmass was for convective development and maintenance.

GOES-R Nearcast:

  • Theta-E fields pretty good for diagnosing where convection was most likely.

GOES-R UAH SatCast/UW Cloud-Top Cooling:

  • UAH CI started out well on one event, but then began to struggle near the boundaries.
  • Still unsure of the utility to issue advanced warnings with the CTC product.  However, could be used for significant weather advisories and pre-warning products.
  • Training/best practices need to be developed that match the use of the CTC and CI products to near-storm environment, as performance varies with varying NSE.
  • There is utility in CI product for situational awareness and cell maintenance with MCSs.

GOES-R PLGM and Lightning Trend Tool:

  • Recommend a “drag me to storm” type interface for the trend tool.
  • Trend tool still not user-friendly, is time consuming to use.
  • Works well for DSS of NWS issued a lightning product/forecast, but not necessarily hail or tornado warnings yet.

OVERALL COMMENTS:

  • The Articulate training modules received a lot of praise.

CONTRIBUTORS:

Greg Stumpf, EWP2013 Week #3 Weekly Coordinator; EWP2013 Operations Coordinator

Travis Smith, EWP2013 Week #3 Backup Coordinator

 

Tags: None

Another Warning…Almost…and Some Other Thoughts…

Ok, so it’s always important to have a good understanding of new products before going gung-ho so to speak.  I thought I’d blog about something different here.  On my warning screen today, I had been looking at the 30 minute accumulation MESH product (or so I had forgotten it was the 30 minute accumulation) and yesterday when we were “in” New York, I had a great deal of success using this product, in conjunction with others.  Initially, I noticed a >1 inch value over northern Andrews County as shown in image 1 below.

Image 1.  MRMS MESH at 2316 UTC.  Notice the >1 values over northern Andrews County.
Image 1. MRMS MESH (30 min accumulation) at 2316 UTC. Notice the >1 inch values over northern Andrews County.

My old warning had expired and I started to prepare a new warning (I was blogging and was late keeping up with things =P ), but then thought it would probably be proper to more closely scrutinize some other data. I switched over to the MRMS reflectivity at -20C and noticed the values were relatively low for this cell at the time, with vales generally below 45dBZ in this same area (image 2).

Image 2.  2316 UTC MRMS reflectivity at -20C.  Notice the relatively low values in northern Andrews County.
Image 2. 2316 UTC MRMS reflectivity at -20C. Notice the relatively low values in northern Andrews County.

So, what did I do next?  I went back to the data I knew best and was most familiar…the all-tilts.  At the nearest time (2318 UTC), I noticed relatively low values of dBZ even near the 20kft level in this area, which corresponded to the MRMS reflectivity at -20C data (image 3).

Image 3.  2318 UTC KMAF 5.1 degree reflectivity.  At this elevation angle, the radar slice was approx 22 kft AGL at this location in northern Andrews County.  Notice generally low dBZ values at this height.
Image 3. 2318 UTC KMAF 5.1 degree reflectivity. At this elevation angle, the radar slice was approx 22 kft AGL at this location in northern Andrews County. Notice generally low dBZ values at this height.

From there I went to my four-panel of MRMS data, which included the 0.5 degree KMAF HSDA (image 4 below).  I had gained familiarity with these products fairly well over the past couple of days and wanted a quick sanity check.

Image 4. 2316 UTC MRMS and  2318 KMAF HSDA data, overlaid with SVR warnings. Upper left - MRMS Merged reflectivity QC Composite, MRMS upper right - reflectivity at -20C, lower right - MRMS MESH, lower left - KMAF 0.5 degree HC.
Image 4. 2316 UTC MRMS and 2318 KMAF HSDA data, overlaid with SVR warnings. Upper left – MRMS Merged reflectivity QC Composite, MRMS upper right – reflectivity at -20C, lower right – MRMS MESH, lower left – KMAF 0.5 degree HC.

Notice in image 4 above, the low MESH values in northern Andrews County at the time.  What?  I thought there were values over 1 inch.  I went back to my warning screen and noticed that I had the 30 minute accumulation of MESH.  I had simply forgotten.  However, this had almost resulted in a faulty warning.  Now, it could have been the HWT environment.  Would I have been so quick to react and careless in an actual operational setting?  Probably not.  However, this does bring to mind the importance of making sure to look at products carefully, especially new products with which the forecaster may not always be familiar.  Yesterday, when we were “in” the BUF office, I had used this product in addition to the MRMS reflectivity at -20C to actually form my polygons, and it had worked pretty well I thought.  But those cells were tracking along nicely.  Today, the cells were moving quite slowly.  So, the 30-min accumulation product may not have been appropriate to use in conjunction with WarnGen.  It was hard to discern any movement or much change in the product since the cells were not tracking along much. That’s why it was hard to determine if the >1 inch MESH had happened recently or much earlier in the 30 minute window.  Lesson learned.  This product may have been fine for SA purposes and perhaps kept on my non-warning screen.  These are things to think about for proper warning strategies.

Kris

 

Tags: None

50dbz Echo Top_MESH_HSDA North Texas

The 50 dbz echo top continues to be one of my favorite parameters for quick and easy interrogation should things get busy. It’s a nice way to see the approximate height of the 50dbz core. Height of the 50dbz core above -20C is also a nice touch to analyze those really big storms that may need more significant hail wording.50_60dbz_2310z

The MESH and HSDA were both doing a good job with this storm. We did not get any hail reports, but that is likely due to the rural locations where the storm was tracking. Seeing the MESH indication of 1 inch hail and the HSDA forecast of large hail added confidence in the warning decision making process and it only took a few seconds to glance at these products. Therefore, there was still time to go through a complete review of base reflectivity and SRM data in all_tilts. The algorithm hail trends reflected what one would expect based on all_tilts base data.MESH_HSDA_2313Z

The mid level rotation track was a nice way to view the life of this storm. Initially there was good mid level rotation, but this rotation begins to weaken with time and the warning was allowed to expire early. The decision to expire the warning early was based primarily on trends within MESH, HSDA, and 50dbz echo tops. But the loss of rotation was icing on the cake for letting the warning expire.  (Jeremy Wesely)

Rotation_Storm_Decay

Jeremy Wesely

Tags: None

Storms Along the SJT/MAF Border

An earlier cell merger helped enhance low-level rotation and produced a reported tornado near Rotan in Fisher County TX in SJTs area.  The mid-level meso would eventually track SW toward Scurry and Mitchell Counties in the MAF forecast area.  The LAPS streamline analysis at 2330 UTC indicated convergence over the northeast portion of the MAF CWA, which appeared to be concentrating higher theta-e values in the location likely allowing the stronger convection to translate west of south.

2330LAPSradar

Tags: None

Damaging Convective Wind Event To Affect Areas From Abilene to Throckmorton

Our first clue was the 2.5KM LAPS Theta-E image with LAPS wind and regional radar overlaid on it at 2245 UTC.2245ZLAPS_Theta-E_Wind*The tornadic Fisher Co storm was propagating southward toward the bullseye of Theta-E over Nolan Co, while the rest of the more linear convection with damaging to destructive winds was moving southeast toward an area of high Theta-E extending from Jones/Shackleford countis into Callahan/Brown counties.

There’s a chance this linear complex splits with the better tornadic part moving southward across western San Angelo’s CWA from Fisher south through Nolan and Coke counties where the OUN-WRF 0-1KM SRH forecasts values by 0200 UTC to increase over 200 m2/s2. Meanwhile, the linear activity moving toward Abilene/Throckmorton is moving into an area of 0-1KM SRH less than a 100 m2/s2 for the most part with likely more of a damaging wind/marginal hail threat.

0200UN_WRFSRH

Here is latest MESH and 60 min MESH swatch from the severe activity.

SJT_MESH*

Tags: None

Some further warnings in the Midland CWA prompted by MRMS and HSDA data…

A few more warnings have resulted in the Midland CWA since I switched over.  I’ll blog about the most recent.  By 2300 UTC I noticed a line of deep convection moving northward through Brewster County, TX.  Unfortunately, this area was on the far southern end of our domain today, so I didn’t get HWT specific products until some of the potentially severe portions of these cells may have already been ongoing.  Anyway, as you’ll notice, MRMS MESH and HSDA data were indicating potentially severe hail with this line of storms between 2250-2258 UTC.  I had to use the KMAF reflectivity data to produce the warning (to see the extent of the line of convection).  Nevertheless, you’ll notice that the MESH was as high as about 2 inches in the extreme southern portion of the image.  HSDA data at the time was showing a line of categorical giant and large hail.

Image 1.  2258 UTC MRMS and HSDA data, overlaid with SVR warnings.  Upper left - MRMS Merged reflectivity QC Composite, MRMS upper right - reflectivity at -20C, lower right - MRMS MESH, lower left - KMAF 0.5 degree HC.
Image 1. 2258 UTC MRMS and HSDA data, overlaid with SVR warnings. Upper left – MRMS Merged reflectivity QC Composite, MRMS upper right – reflectivity at -20C, lower right – MRMS MESH, lower left – KMAF 0.5 degree HC.

These data prompted me to issue a warning, which went out at 2304 UTC.  The following image shows the resulting polygon.

Image 2.  2259 UTC KMAF 0.5 reflectivity and resulting warning polygon.
Image 2. 2259 UTC KMAF 0.5 reflectivity and resulting warning polygon.

Any reports so far?  No…I’m not real sure to expect any either, but we’ll see.

Kris

Tags: None

LAPS CAPE Field for Outflow

LAPS CAPE fields rather quickly caught onto the idea that the northern LBB/southern AMA area would be dominated by cooler, more stable surface air as a result of outflow from the earlier supercells southeast of Lubbock.  The most instability is limited to the far western portions of the CWAs.

2245LAPSCAPE

Tags: None

Storm close to radar_CTC_Storm Initiation_50dbz echo top_MESH_HSDA

The MRMS products were a big help in this case being able to look at multiple radars since my storm was right near the radar as depicted in the image below.  (Wesely)

BASE_REFLECTIVITY

The Height of the 50dbz core went above 30,000ft agl, which was a trigger for issuing the warning. The 4 panel product below made the process easier in two ways. First the storm was near the radar and so having a 50 dbz echo core height product that took into account multiple radars was extremely helpful as the local office radar was unable to see the top of the storm. In addition, you could just look at the one screen and see the 50 and 60 dbz core height quickly without having to look through multiple slices of an adjacent offices radar to find the height of these cores.

50_60dbz_2012z

The MESH and HSDA algorithms below both provided lead time for the quarter size hail that fell 11 miles north of Amarillo at 2236Z. The multi radar aspect of the MESH algorithm also made it easier to use than having to look at an adjacent radar to see the top of the storm. The HSDA is not multi radar and thus there were some limitations with this algorithm for the storm near Amarillo due to the close proximity of the storm to the radar.mesh_2212ZHSDA

Cloud Top Cooling (CTC) did a nice job of showing that convection was really trying to break out in Amarillo’s County Warning Area (CWA). There were numerous cloud top cooling signatures beginning at 20z. The first 50dbz echo reached the ground at 2146z. So it took awhile but eventually the storms overcame the cap and became severe (See image of CTC below).

CTC2115z

Jeremy Wesely

 

Tags: None

HSDA Detections within non-uniform beam filling

Some HCA detections in the east (right most storm) are occurring within areas of non-uniform beam filling from the first storm (and most likely the second storm).  Forecasters should be wary of identifications within these regions and shows the need for investigation of the base variables in addition to using the algorithms for diagnosis of storm severity.

nubf_hca nubf_base

 

–KO

Tags: None