John Esterheld and I were able to present our research project today at the NWC in Norman, OK, and are now ready to publicly share our results. We appreciate the time spent by those who were able to attend. Here is a link to our PowerPoint presentation file:
http://americium.gcn.ou.edu/presentation.ppt
This file is best viewed as a slideshow in Microsoft Powerpoint, as there are animations/highlights meant to be viewed in sequence and the file does not work correctly with OpenOffice.
We are now moving on to the task of writing this up for peer review...questions, comments, etc. are welcome.
I think Rich, Chuck and Roger brought up good points on the ability to use this in an operational sense (beyond using some gut feeling).
It will be difficult to use this operationally, but I think with a little more research into this feature experienced forecasters will be able to take advantage of it, particularly from a nowcasting perspective. What makes it so tough to utilize is the fact that it is highly sensitive to even fairly small variations in low-level flow and storm motion, so there's a danger of the forecaster being overwhelmed by information overload, or on the other hand to overanalyzing sparse data and "seeing" features that do not exist.
Институт Теоретической Физики имени Л.Д. Ландау::
Kamenshchik, Relativistic hodograph equation for a two-dimensional U. Alam, V. Sahni, A.A. Starobinsky, The case for dynamical dark energy revisited,
http://www.itp.ac.ru/itp_lite/index.php?specinc=one_p&toshow=/2004/astro&translit=noHOME |
M. Adler (Brandeis, USA) - PDE's for joint distributions of Dyson ::
equations are hyperbolic and can be integrated through the so called hodograph transform. . F. Magri (Milano, Italy) - The Kowalevski top revisited
http://www-math.univ-poitiers.fr/PvM/abstract.htmlHOME |
I think one area that would really help using the techniques you did, combining OK Mesonet/profiler/soundings, is looking at strange, or atypical, tornado events. Several events in OK this year where quasi-linear (for lack of a better classification) or shallow events produced not significant (i.e., (E)F-2+) tornadoes, but tornadoes nonetheless. Most were extremely short lived, came from fairly "benign" velocity signatures but still produced situations where people could have been killed or injured. Since most came from days without uber-instability I think looking at the kinematic fields as you guys did might help more than using some bulk parameter. The nice thing about the events is that there is quality verification of the tornadoes done by Patrick Marsh, Kevin Manross, Rick Smith and me and a number of radar resources to use (88D, CASA and PAR). I think possibly using your "critical angle" idea might help in figuring out why these storms produced, why they produced when they produced and possibly opening up some avenues in forecasting their occurance.
It would certainly be interesting to examine some of these unusual tornado events more closely to see if this feature is evident where high-resolution observations are available. In my experience, it is almost universally present in the outbreak cases, but we have also found plentiful anecdotal evidence that it is present in more isolated cases such as the Seymour, TX, warm-front tornado earlier this year, or the Harper County tornado-fest May 12th, 2004.
Another avenue that I think is promising is the potential ability to distinguish between tornadic/non-tornadic storms which co-exist in the same environment (see the supplemental slide from March 28th of this year with the three storm motions plotted on it...I'm sure you can guess which one failed to produce), as well as to potentially determine when a storm will become tornadic (e.g., May 9, 2003, which we included as part of our presentation).
Don, you and John made an excellent presentation. Had some problems with combining Haviland and VNX's VAD profile for Greensburg, but that's neither here nor there given it was just an example. I think Rich, Chuck and Roger brought up good points on the ability to use this in an operational sense (beyond using some gut feeling). I think one area that would really help using the techniques you did, combining OK Mesonet/profiler/soundings, is looking at strange, or atypical, tornado events. Several events in OK this year where quasi-linear (for lack of a better classification) or shallow events produced not significant (i.e., (E)F-2+) tornadoes, but tornadoes nonetheless. Most were extremely short lived, came from fairly "benign" velocity signatures but still produced situations where people could have been killed or injured. Since most came from days without uber-instability I think looking at the kinematic fields as you guys did might help more than using some bulk parameter. The nice thing about the events is that there is quality verification of the tornadoes done by Patrick Marsh, Kevin Manross, Rick Smith and me and a number of radar resources to use (88D, CASA and PAR). I think possibly using your "critical angle" idea might help in figuring out why these storms produced, why they produced when they produced and possibly opening up some avenues in forecasting their occurance. A Posteriori Error Estimation for Discontinuous Galerkin ::
File Format: PDF/Adobe Acrobat - View as HTMLrevisits. the topic of a posteriori. error estimation. of prescribed. functionals __olution via hodograph. transformation. 2 .4 Po_.teriort
http://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20000102376_2000148099.pdfHOME |
Don, you and John made an excellent presentation. Had some problems with combining Haviland and VNX's VAD profile for Greensburg, but that's neither here nor there given it was just an example.
Kiel,
Thanks for your kind words, we appreciate your attendance.
Perhaps I wasn't clear as to what we did to plot up the profile for the Greensburg storm, as we simply used the Haviland surface observation (not the profiler), and combined that with the Vance AFB VAD. We didn't combine two upper level data sources into one, which I think you may have interpreted that as.
--Don
More evidence of the importance of the near surface layer which can often have significant variability over short distances. Makes you wonder how much our sampling spreads the histogram/box plots. Now we need to 100x the sample size and include thermodynamic data =)
More evidence of the importance of the near surface layer which can often have significant variability over short distances. Makes you wonder how much our sampling spreads the histogram/box plots. Now we need to 100x the sample size and include thermodynamic data =)
I think better sampling would yield narrower spreads for the boxplots and histograms. For instance, one of our best sigtor cases was jettisoned because the profiler data were QA'd out at the time of the tornado (even though we regard the data as perfectly realistic), and another case comes to mind where a boundary passed through the only profiler that fit these criteria a mere hour *before* the time of the tornado, and we wound up having to use another profiler which did not work out at all. Maybe we should all outfit our chase vehicles with sodars and roam the plains with a high-density mobile sodarnet. ;)
As far as thermodynamics are concerned, if given the choice I'd choose higher CAPE, but am unconvinced that it really plays as much of a role as is commonly thought, CAPE vs. shear plots notwithstanding. My suspicion is that CAPE and shear are simply inversely related in general (think about all the high-helicity days in the winter), and that the presence of strong shear in the correct configuration is sufficient to produce a violent tornado given a surface-based supercell regardless of the strength of the instability, such as on Feb. 28th of this year. It is the rare instances when both are present in plentiful quantities when things really get out of control (e.g., May 4/5 of this year).
 Red Hat's Rough Recovery From CFO Exit
 Windows Live Finds a New, Pre-installed Home |
