Toward A Radar-Based Climatology Of Mesocyclones

1
Toward A Radar-Based Climatology Of Mesocyclones

• 2nd Conference on Severe Storms in Europe
• Prague, CR

John T. Snow, Kevin M. McGrath, and Thomas A.
Jones University of Oklahoma Norman, OK
2
Objectives

• Long-term Produce a climatology of mesocyclones
  in the southern Great Plains
• Immediate Assess feasibility of constructing
  such a climatology using data from the national
  network of WSR-88D radars

3
Procedure

• Use high-resolution data from WSR-88D network
• Process these data using a realization of the
  Mesocyclone Detection Algorithm (MDA) developed
  at NOAA National Severe Storms Laboratory ?
  climatology of mesocyclone detections derived
  from this particular algorithm
• Improve the quality of the detection data set by
  identifying and removing spurious detections ?
  climatology based on filtered data set
• Associate mesocyclone detections with
  mesocyclones in nature ? severe weather reports

4
Radar Data

• High-resolution (Level II) data have been
  acquired from multiple Southern Plains radars
  under the auspices of the Collaborative Radar
  Acquisition Field Test (CRAFT) project

• Convective cases from 2000 and 2001 from the
  initial set of six radars (KAMA, KFWS, KINX,
  KLBB, KSRX, and KTLX) have been processed using
  the MDA additional radars available for 2002
• Approximately 2500 hours of data have been
  processed from each radar of the initial set of
  six radars, 80 from 2001

5
Algorithm Output

• Location of center of mesocyclone ? analyze,
  display on Geographic Information System,
  associate with severe weather reports
• Many other parameters indicating strength, size,
  intensity of shear, etc of the mesocyclone ?
  basis for filtering detections
• Companion algorithms provide additional
  information about parent thunderstorm
• Point Developed to support operational
  forecasting, not climatological study

6
Challenges

• Large amount of data requires an almost fully
  automated procedure
• High number of weak detections (Mesocyclone
  Strength Rank 0) tends to obscure the stronger
  and more significant detections
• Obvious spurious detections caused radar
  characteristics and algorithm limitations
• Ground clutter
• Anomalous propagation
• Incorrectly de-aliased velocity data

7
Filtering Techniques

• Remove false MDA detections that meet any of
  the following criteria
• Located within 5 km of the radar
• Located at the maximum unambiguous velocity range
• Weak in intensity (Meso. Strength Rank 0)
• Detected in clear air mode (VCP 31 or 32)
• Not associable with a SCIT-defined storm cell at
  time of detection

Initial Filtering
SCIT Filtering
8
Determining SCIT Filter Radius
Correlation of Mesocyclone Low-level Rot. Vel.
And SCIT Derived Storm-cell VIL as a Function of
Separation Distance Between Centroids
of KTLX Mesocyclone Detections Retained as a
Function of SCIT Filter Search Radii
9
Example of SCIT Filtering
KAMA, 20010502 15Z 20010504 0Z
MDA detections, post-initial filtering. Note
region of high ranking, false detections.
Mesocyclone track
KAMA
KAMA
MDA detections remaining after passage through
the SCIT filter (10 km circular window). Meso
track now much clearer.
10
Unfiltered 2000 and 2001 KTLX Detections
N 256,345
Mesocyclone Detections
Density of Mesocyclone Detections
11
True 2000 and 2001 KTLX Detections
N 18,788
SCIT-Filtered? Mesocyclones
Density of SCIT-Filtered? Detections
? Using a circular window of 10 km.
12
True 2000 and 2001 KTLX Detections
Equal Area Range Bins Histogram
Azimuth Histogram
13
Number of MDA Detections
Radar Non-filtered Detections Retained after Initial Filtering? Retained after SCIT Filtering (10 km circular window)
KAMA 233,558 8.9 (20,837) 5.4 (12,588)
KFWS 266,694 9.0 (23,941) 6.5 (17,206)
KINX 367,199 6.4 (23,649) 4.3 (15,927)
KLBB 163,374 8.5 (13,812) 5.0 (8,224)
KSRX 302,812 8.0 (24,126) 5.5 (16,573)
KTLX 256,345 10.5 (26,856) 7.3 (18,788)
? Removed detections with range ? 5 km, range
equal to maximum unambiguous velocity range, MSr
0, or those detected in VCP 31 or 32.
14
True Detections Using a 10 km search window
KAMA
KFWS
KAMA
KINX
KTLX
KLBB
KFWS
KINX
15
True Detections Using a 10 km search window
KAMA
KINX
KTLX
KINX
KSRX
KFWS
16
Density of True Detections Using a 10 km search
window
KAMA
KFWS
KAMA
KTLX
KLBB
KLBB
KFWS
17
Density of True Detections Using a 10 km search
window
KAMA
KTLX
KINX
KFWS
KFWS
KINX
KFWS
KSRX
18
MesocyclonesCyclonic and Anticyclonic

• Data are processed twice, once to detect cyclonic
  mesocyclones, a second time to detect
  anticyclonic mesocyclones same filtering
  technique used each to remove false detections

19
Cyclonic Detections after initial SCIT
filtering (10km window) 1215
20
Anticyclonic Detections after initial SCIT
filtering (10km window) 851
21
(No Transcript)
22
May 5 6, 2002 Mesocyclones
Cyclonic Detections after initial filtering 4601
after SCIT filtering (10km window) 3139
23
(No Transcript)
24
May 5 6, 2002 Mesoanticyclones
Anticyclonic Detections after initial filtering
4055 after SCIT filtering (10km window) 2617
25
Associating Detections With Tornadoes

• Use GIS system to associate reported tornadoes
  temporally and spatially with true detections
  of mesocyclones time ? 30, -10 minutes, space ?
  w/i 10 km
• N.B. Some tornado reports not associated with a
  pre-SCIT filtered mesocyclone detection SCIT
  filtering of detections resulted in a few
  additional tornadoes not being associated with a
  true mesocyclone detection

26
2000 Tornadic Detections
KINX
KTLX
KSRX
KAMA
KLBB
KFWS
27
2001 Tornadic Detections
KINX
KSRX
KTLX
KAMA
KLBB
KFWS
28
A Few Early Conclusions
29

• A large percentage of MDA detections are spurious
  ? probably real shear regions, but not
  mesocyclones study provides different
  perspective on performance of the operational
  algorithm
• The quality of a mesocyclone detection data set
  can be significantly improved using rather simple
  filtering techniques
• Results for an area dependent on how radar is
  operated in that region ? national network, but
  each radar is under local control
• Surprising number of anticyclonic events ?
  algorithm artifact?
• Very small percentage of tornadic mesocyclones ?
  long period 10 years?, 20 years? required to
  develop a climatology with high degree of
  certainty

30
Work Underway
31

• Processing of 2002 data continues
• Expanding the study to include KDDC, KFDR, KICT,
  and KVNX
• Developing filtering techniques that require less
  human interaction. Specifically, filtering of
  the concentration of detections at so-called
  first trip rings
• Exploring use of existing data set for evaluating
  of skill in tornadic forecast parameters
• Exploring how to group multiple individual
  detections into families which represent single
  mesocyclones
• Reviewing nature of algorithms to identify
  possibility that some, perhaps many anticyclonic
  detections are artifacts of a cyclonic detection
  (works other way, too!)

32
Acknowledgements

• Don Burgess, NSSL
• Kelvin Droegemeier, CAPS
• Jason Levit, CAPS
• Greg Stumpf, NSSL
• Andy White, School of Meteorology, OU
• Oklahoma NASA Space Grant Consortium
• NOAA Warning Decision Training Branch
• Point True Oklahoma Weather Center project,
  would not have been possible without assistance,
  collaboration by many folks in different
  organizations

33
Contact Information

• John T. Snow
• College of Geosciences
• University of Oklahoma
• 100 E. Boyd, Suite 710
• Norman, OK 73019 USA
• Telephone [phone]
• FAX [phone]
• E-mail jsnow_at_ou.edu
• Project URL http//mesocyclone.ou.edu

34
(No Transcript)
35
False 2000 and 2001 KTLX Detections
N 8,067
SCIT Filtered? Mesocyclones
Density of SCIT Filtered? Detections
? Using a circular window of 10 km.
36
False Detections Using a 10 km search window
KAMA
KTLX
KFWS
KLBB
37
False Detections Using a 10 km search window
KINX
KTLX
KFWS
KSRX
38
Density of False Detections Using a 10 km
search window
KAMA
KTLX
KFWS
KAMA
KFWS
KLBB
39
Density of False Detections Using a 10 km
search window
KINX
KTLX
KFWS
KAMA
KSRX
KFWS
KINX