TOP CWA Golf Ball size Hail Study by suchenfz

VIEWS: 21 PAGES: 32

									TOP CWA Golf Ball Size Hail Study




    Bill Gargan WFO TOP, KS
  Why I conducted this research?
• To find techniques to help forecasters, while
  analyzing storms on radar, to generate more
  accurate warnings and severe weather
  statements for golf ball size hail.
• Golf ball size hail can cause considerably
  more damage than quarter size hail.
                    DATA
• Archived reflectivity (Zr) and storm relative
  velocity (SRM) data was obtained from both
  the KTWX and KEAX radars. Archived
  radiosonde data from the KTOP upper air
  observing site was used to obtain the height
  of both the melting and -20 degree C levels.
• 174 golf ball size hail reports were obtained
  from Storm Data across the TOP County
  Warning Area from 1999 to 2008.
  The following components analyzed
     from Zr and radiosonde data
• Maximum Zr at -20 degree C. (From HDA, most severe
  hail growth occurs with temperatures near -20C or
  colder. )
• Maximum Zr value and corresponding AGL height.
• 2Vr (rotation).
• Storm Top Divergence.
• 50 dBZ Height vs the melting level height. (Donovan 1”)
• Percentage of storms with identifiable BWER
• Percentage of storms with three-body scatter
  signature.
    Data collection methodology
• The radar data was analyzed up to 2 volume
  scans before the time of a golf ball size hail
  report and up to one volume scan after the
  time of a golf ball size hail report. For all the
  radar parameters, I looked at the maximum
  values within the time frame.
• Some observation errors are inherent to this
  study based on the over and under estimation
  of hail size by the storm spotters, and potential
  errors in the time of the report.
  The following components analyzed
     from Zr and radiosonde data
• Maximum Zr at -20 degree C
• Maximum Zr value and corresponding AGL
  height.
• 2Vr (rotation).
• Storm Top Divergence vs the melting level height.
• 50 dBZ Height vs the melting level height.
• Percentage of storms with identifiable BWER
• Percentage of storms with three-body scatter
  signature.
    The following components analyzed
       from Zr and radiosonde data
• Maximum Zr at -20 degree C
• Maximum Zr value and
  corresponding AGL height.
•   2Vr (rotation).
•   Storm Top Divergence.
•   50 dBZ Height vs the melting level height.
•   Percentage of storms with identifiable BWER
•   Percentage of storms with three-body scatter
    signature.
    The following components analyzed
       from Zr and radiosonde data
• Maximum Zr at -20 degree C.
• Maximum Zr value and corresponding AGL
  height.
• 2Vr (rotation).
•   Storm Top Divergence vs the melting level height.
•   50 dBZ Height vs the melting level height.
•   Percentage of storms with identifiable BWER
•   Percentage of storms with three-body scatter
    signature.
Mean 2Vr = 57 KTS
 The following components analyzed
    from Zr and radiosonde data
• Maximum Zr at -20 degree C
• Maximum Zr value and corresponding AGL
  height.
• 2Vr (rotation).
• Storm Top Divergence.
• 50 dBZ Height vs the melting level height.
• Percentage of storms with identifiable BWER
• Percentage of storms with three-body scatter
  signature.
     Storm Top Divergence (STD)
• Divergence was observed at or near the top of
  the storm using SRM.
• STD = |∇(Vro,-Vri)|= |Vro – ( -Vri)|
Mean STP X = 109 KTS, SDV= +/- 25 KTS, X-STD=(84 KTS, 134 KTS)
    The following components analyzed
       from Zr and radiosonde data
•   Maximum Zr at -20 degree C
•   Maximum Zr value and corresponding AGL height.
•   2Vr (rotation).
•   Storm Top Divergence.
• 50 dBZ Height vs the melting level
  height.
• Percentage of storms with identifiable BWER
• Percentage of storms with three-body scatter
  signature.
Mean X = 39.1, STD = +/-5.9, 2STD = +/-11.8, X-STD = (45,33), X-2STD = (51,27)
  50 dBZ Height vs. Melting Level
• Recorded the approximate 50 dBz height for
  each storm where Golf Ball Size hail was
  observed.
• Plotted the 50 dBz height vs the height of the
  melting level.
    The following components analyzed
       from Zr and radiosonde data
•   Maximum Zr at -20 degree C
•   Maximum Zr value and corresponding AGL height.
•   2Vr (rotation).
•   Storm Top Divergence.
•   50 dBZ Height vs the melting level height.
• Percentage of storms with
  identifiable BWER
• Percentage of storms with three-
  body scatter signature.
      Examining Storm Structure
• I noted all the storms that had either a Three
  Body Scatter Spike (TBSS) and/or Bounded
  Weak Echo Region (BWER).
Example of a BWER
               BWER result
• 10 percent of the storms that produced Golf
  Ball size hail had a definable BWER’s.
• Almost every storm examined had a WER.
• BWER detection is often dependent on
  distance from radar and beamwidth
TBSS example
               TBSS results
• 15 percent of the storms that produced Golf
  Ball size Hail had a definable TBSS.
• However, TBSS cannot be seen easily for many
  azimuths of the storm, relative to the RDA
  location, i.e. TBSS could be difficult to see
  where it would be placed in an area already
  covered by precipitation.
                Final Result
• The best correlation was noted between the 0
  deg C height and 50 dBZ height. A polynomial
  can be fitted to provide a 50 dBZ height given
  the observed (SND) or forecasted (RUC) 0 deg
  C height.
             Another Finding
• Observing the height of the 50 dBZ core along
  with STD may provide you with a technique to
  rule out a storm’s potential for producing golf
  ball size hail.
          STD vs 50 dBZ Height
• Less than 3 percent of all Golf Ball size hail
  reports occurred with 50 dBZ heights less than
  30,000 feet and STD less than 80 KTS.
• Less than 6 percent of all Golf Ball size hail
  reports occurred with 50 dBz heights less than
  30,000 feet.
                 Conclusion
• The 50 dBz height vs the melting level gives
  the best correlation from this study may and
  help the radar operator to determine when a
  severe storm may produce Golf Ball size hail.
• Increase confidence in determining when a
  storm is capable of, or unlikely, to produce golf
  ball size hail and may lead to more accurate
  hail size inclusion in SVS’s.
• Point soundings from the time and location of
  observed golf ball size hail from using NARR
  grib data would have given more accurate
  melting levels.
               Future Research
• To use NARR grib data, instead of the 12Z or 00Z TOP
  upper air data, to determine melting height for both
  the location and time of golf ball size hail report.
• Use the NARR grib data to determine how well the
  Maximum Expected Hail Size (MEHS) algorithm
  performs.
• Currently examining Radar and Environmental
  Conditions associated with Baseball size hail in the
  NWS Topeka County Warning Area
• Also, one of several co-PI’s working on a four inch and
  larger hail study across the plains states.

								
To top