1. Attendees: Matt Parker (NCSU), Jason Davis (NCSU), Keith Sherburn (NCSU), Mike Cammarata (CAE), Frank Alsheimer (CHS), Doug Berry (CHS), Wendy Moen (CHS), Trisha Palmer (FFC), Justin Lane (GSP), Bob Frederick (MHX), Steve Keighton (RNK)
2. Critical Action Item: As soon as you can, please check the Google Earth SPC mesoanalysis data archive and make sure that all of the data that you need are there. We have found that some of the cases are missing several hours of data.
3. Most of the call surrounded the presentation of the Case Study Methodology. There were several aspects of the methodology that were discussed in detail:
a. Appropriate temporal resolution of the radar analysis discussion. GSP’s template case study included detailed analysis of each event beginning about one hour before the occurrence of severe weather. However, the radar analysis section was much too long. It was decided that we should focus on the 15-minute window prior to the event for our detailed discussion of radar data. This will help keep things more manageable. However, time series plots of rotational shear, rotational velocity, and NROT should be for a longer duration in order to reflect the trends in that data (see next section).
b. It was decided that rotational velocity and rotational shear should be calculated where applicable and plotted in a time series format. There was some discussion over the “proper” method for calculating these parameters, as we should all be using the same method. The method that we will use is a.) Locate the updraft, b.) Locate the maximum inbound velocity associated with the updraft, c.) Locate the maximum outbound velocity associated with the updraft. d.) As long as the outbound/inbound pair are more rotational than convergent/divergent (subjectively determined), perform the shear/rotational calculations. However, there will undoubtedly be situations that do not conform to this simplistic methodology. It might be a good idea to share these situations with the HSLC group for discussion.
c. NROT will be analyzed for its utility in HSLC events. Time series plots of NROT should be presented alongside plots of shear/rotational velocity. A manuscript describing NSSL’s LLSD algorithm for computing azimuthal shear (the basis for NROT) was shared. A presentation shedding light on NROT as well as other GR2 algorithms (from Scott Lincoln, NWS/RFC Slidell) was also shared. It has been placed on the E.R. ftp server:
You do need a NOAA e-mail username and password to access the site, so if the NCSU folks are interested in seeing it, I’ll have to make other arrangements to get them the file.
Once you’ve logged in, navigate to “Folders” (upper left side of the page), then “Share” “GSP” “CSTAR” “GR2”.
d. Tornado Warning False Alarms- It was pointed out that modifying RUC soundings in the vicinity of cells prompting false alarm TORs was equally important to radar analysis. Therefore this step was added to the methodology. However, due to workload concerns, it was decided that analysis of these cells would be limited to 2-5 for each case.
e. There was an overall concern voiced about the workload of the case studies, especially in light of the encroaching convective season. All I can say is recruit as many folks as you can to help out, and complete as much work as you can by November. If you are able to complete only one case study during that time, you have still made a huge contribution to the project.
4. The rough draft of the radar analysis section can be accessed on the E.R. ftp server under “Share” “GSP” “CSTAR” Review.docx.
Don’t be scared off by the length. It has not been edited yet based on reviewers’ comments and the changes that have been made to the methodology. It’s probably twice as long as it needs to be. However, this is the basic format that we will follow.
5. FFC has already delivered the database of null cases to the NCSU folks. Thanks to Trisha for taking the lead on this. NCSU will be making a null case data request to SPC soon. They will also incorporate the request for any missing data for our case studies at the same time. Again, please make sure that all of the data that you require for your case studies is on the N.C. State server. E-mail Matt, Jason, and Keith with your data request.
6. Keith gave an update on his latest work. His parameter study is now focused on discriminating between “wind days” and “tornado days” and among convective modes. Preliminary results seem to indicate that, contrary to conventional wisdom, instability may be an important discriminator of convective mode for HSLC events. Also, the orientation of the shear relative to surface boundaries seems to be important in “wind days” vs. “tornado days,” with tornadoes being favored when there is more of a “cross boundary” component to the shear vector. Keith’s power point presentation can be found on the E.R. ftp server under “Share” “GSP” “CSTAR” Keith_120405.
7. Jason’s latest work involves creating time series plots of radar “azimuthal shear” for events within our case studies. This is an algorithm from NSSL’s WDSS-II software. (It is used to create the “Rotation Tracks” products that I’m sure we’ve all examined from time to time). It is similar, but not the same as NROT (actually the algorithm presented in the manuscript in 3c. above is probably more descriptive of the WDSS-II algorithm than of NROT). Jason will continue to plot these values. However, it was agreed that we will all still need to calculate the more traditional parameters (Vr and rotational shear) for our case studies since these algorithms are not available operationally. Jason’s presentation from the call can be found on the E.R. ftp server under “Share” “GSP” “CSTAR” Jason_120405.
Here is Jason’s explanation of what he is working on and what you are viewing in the presentation…
“For the velocity signature portion of the radar climatology, I have chosen to focus on studying azimuthal shear over the lifetime of numerous tornadic and nontornadic circulations (including but not limited to false alarm tornado warnings). My plan is to study at least a few, if not more, of our HSLC cases, find the circulations associated with tornado reports, and identify other areas of rotation on radar that appear to be of interest but did not produce a tornado. I would then use a script to track the areas of associated azimuthal shear maxima over the lifetime of the circulation at multiple radar elevation angles. Azimuthal shear is calculated by NSSL’s WDSS-II program, and is what is shown in the “rotation tracks” maps you may have seen. It is calculated by an LLSD method that uses more than just two velocity bins, and is explained very well in Scott Lincoln’s PowerPoint presentation that was recently sent out. The calculation is very similar to that used in NROT, except that NROT divides the shear value by a threshold based on range from the radar, since azimuthal shear will tend to decrease with range due to radar sampling. The thresholds used in NROT were based on comparing shear values associated with tornadoes of the same EF-rating that were located at different ranges from the radar (see Trisha’s e-mail from Mike Gibson). Instead of doing this, I will probably be using an azimuthal shear range-correction method recently developed by NSSL, also mentioned in that e-mail, and explained in this conference paper (http://ams.confex.com/ams/91Annual/webprogram/Manuscript/Paper184514/ams_paper_short.pdf). I should be able to translate my results for azimuthal shear into something similar to NROT, however. I plan to use azimuthal shear rather than v-r shear or NROT because this would allow me to automate the collection process and minimize manual intervention, enabling me to build up a large population of circulations. Once I have this population, I can look at things like thresholds (and possibly associated PODs/FARs), trends, depths of the circulations, potential lead time, composite timeseries of azimuthal shear, and make comparisons between different convective modes and between tornadic and nontornadic circulations. Even if the actual values of azimuthal shear may not have much operational utility, trends in azimuthal shear and information about the depth of the circulations would correspond to what the forecaster sees when looking at the circulation in SRV data or NROT. One difficulty is that gust fronts can also show up in azimuthal shear/NROT as a linear signature, which I will want to distinguish from a rotation signature.
The attached PowerPoint slides show some examples of timeseries plots of azimuthal shear for some tornadic and non-tornadic circulations at multiple elevation angles, as well as the maximum in azimuthal shear in the 0-2 km layer AGL (takes maximum shear value out of the elevation scans that are available in this layer). I have not applied any range correction, so I need to wait before making comparisons between the cases, as they occurred at varying ranges from the radar. The Newberry, SC tornado on slide 3 is probably the most interesting case, as it shows a spike in azimuthal shear at the 1.3 degree and 0.9 degree elevation angles that precedes the tornado by 9 minutes. The other cases do not look as nice, but I’ll have to wait and see what range correction tells me, and continue to look at more and more cases. Let me know if you have any feedback/ideas.”