Providing CG Lightning Guidance

1
Anticipating Cloud-to-Ground (CG)
Lightning Utilizing Reflectivity Data from the
WSR-88D.
Pete Wolf, SOO National Weather
Service Jacksonville, Florida
Image from NOAA
2
Should We Do More?

• CG lightning as weather-related killer2nd to
  floods.
• CG lightning among leading weather-related causes
  of property damage, and most frequent damage
  claim.
• 2 billion annually in U.S. alone.
• Lightning is the most frequent threat to
  life/property.
• Our current effort is primarily focused on public
  awareness.
• 30/30 rule, Take Cover if you Hear Thunder,
  Lightning Awareness Week, etc.
• In real-time, we do more to alert people of
  penny-sized hail than CG lightning.
• More is not done due to belief that all t-storms
  produce deadly lightning.

3
Is CG Lightning Predictable (or at least
Nowcast-able)

• A study on CG lightning initiation radar clues at
  the Kennedy Space Center (KSC) determined that
  skill in anticipating CG lightning was
  achievable.
• The study involved about 40 thunderstorms around
  the KSC area.
• Results showed the best correlation when the
  40dBZ core reached at least the -10C
  heightyielding a POD around 0.85, an average
  lead time of 7.5 minutes, and a low FAR.

4
WFO Jacksonville FL Study

• A more ambitious study was completed in 2005,
  involving 420 convective cells across northern
  FL, southern GA and other areas of the southern
  U.S. The study included a mix of single cell,
  multi-cell, and supercell events occurring during
  various times of the year (a few cases were in
  prior years).
• Nearly 320 cells for CG initiation study, 109 not
  producing CG lightning.
• Additional 109 cells producing numerous CG
  strikes after initially producing
  isolated/scattered strikes
• The study compared storms producing
• No CG strikes (that had 40dBZ core up to at least
  10 Kft, with an echo top up to at least 22 Kft)
• Isolated/Scattered strikes (defined as 1-10 CG
  strikes per 5-minute period)
• Numerous/Excessive strikes (defined as gt 10 CG
  strikes per 5-minute period)

5
WFO Jacksonville FL Study

• KSC Study
• Used Environ. -10C level
• Used 40 dBZ echo height
• JAX Study
• Used Updraft -10C level
• Used 40 dBZ echo height

Updr -10C Envir -10C
6
WFO Jacksonville Findings

• Due to a number of issues, a probabilistic
  approach to the results was taken.
• Radar issues (e.g. interpolation, calibration,
  etc).
• NLDN output accuracy issues.
• Specific values issue (is there a difference
  between 40 dBZ and 38 dBZ or 42 dBZ?)

7
WFO Jacksonville Findings
The results suggested skill in producing
probabilistic guidance based on the 40 dBZ echo
height PROBABILITY OF No
CG CG Nmrs CG Lgtng Lgtng
Strikes 40dBZ height lt -10C level 8
kft 100 0 0 40dBZ height -10C level 5 to
7 kft 87 13 0 40dBZ height -10C level 2
to 4 kft ( -6/-7C level) 67 33 0 40dBZ height
-10C level /- 1 kft (apprch/exceed
-10C) 23 77 0 40dBZ height -10C level 2 to
4 kft 22 78 0 40dBZ height -10C level 5
to 7 kft 0 100 18 40dBZ height -10C level
8 to 10 kft 5 95 41 40dBZ height gt -10C
level 11 to 14 kft 1 99 75 40dBZ height gt
-10C level gt 14 kft 0 100 92 (51)
Note skill in predicting CG strikes when 40dBZ
height reaches -10C height in updraft. Note skill
in predicting numerous CG strikes when 40 dBZ
height gt -10C height 10kft. Note CG lightning
is unlikely when the 40dBZ height is no higher
than -10C level - 5kft The probability patterns
above suggest probabilistic guidance can be
generated for CG lightning.
8
WFO Jacksonville FindingsUpdate
Updated results through mid 2006, incorporating a
total of 835 convective cells
PROBABILITY OF No CG CG Nmrs CG
Lgtng Lgtng Strikes 40dBZ height lt -10C
level 8 kft 100 0 0 40dBZ height -10C
level 5 to 7 kft 92 8 0 40dBZ height
-10C level 2 to 4 kft ( -6/-7C
level) 79 21 0 40dBZ height -10C level /- 1
kft (apprch/exceed -10C) 39 61 0 40dBZ height
-10C level 2 to 4 kft 16 84 0 40dBZ
height -10C level 5 to 7 kft 1 99 11 40dB
Z height -10C level 8 to 10
kft 2 98 35 40dBZ height gt -10C level 11
to 14 kft 2 98 45 40dBZ height gt -10C level
gt 14 kft 0 100 92 (57) Note
skill in predicting CG strikes when 40dBZ height
reaches -10C height in updraft. Note skill in
predicting numerous CG strikes when 40 dBZ height
gt -10C height 10kft. Note CG lightning is
unlikely when the 40dBZ height is no higher than
-10C level - 5kft The probability patterns
change little, again suggesting probabilistic
guidance can be generated for CG lightning.
9
Failure Areas

• This concept does not work well for
• Lengthy anvils, that may not have 40dBZ echo, yet
  can yield bolts from the blue.
• Widespread MCS stratiform regions.
• It is possible to relate these failures to a
  sufficiently high 40dBZ core within the upstream
  (storm-relative) updraft region.
• Manual intervention needed to account for these
  failure areas.

10
CASE EXAMPLES

• Lets take a look at a few cases, to demonstrate
  what the results suggestan ability to anticipate
  occurrence and amount of CG lightning, based on
  WSR-88D reflectivity data.
• April 22, 2005 Supercell/bow echo case
• June 27, 2005 Pulse storm case
• The following are 5-min CG lightning displays
  (from NLDN), with CG lightning probabilities
  overlaid (numerous strikegt10 per 5
  minprobabilities in parentheses).

Very High Probability90 (80) High
Probability60-85 (50-75) Moderate
Probability30-50 (25-45) Low Probability5-25
(5-20)
11
PROBABILITIES USED IN EXAMPLES

• 40dBZ height relative Prob of CG lightning Prob
  of numerous strikes
• to updraft -10C level (1 per 5 min period) (gt 10
  per 5 min period)
• -8 kft
• -7 kft 5
• -6 kft 10
• -5 kft 15
• -4 kft 20
• -3 kft 30
• -2 kft 40
• -1 kft 50
• 0 kft 60
• 1 kft 70
• 2 kft 75
• 3 kft 80
• 4 kft 85 5
• 5 kft 90 10
• 6 kft 95 15
• 7 kft 95 20
• 8 kft 95 25

12
1640z March 22, 2005 Part I
13
1645z
1645z
14
1655z
1655
15
1700z
1700z
16
1705z
1705z
17
1710z
1710z
18
1715z
1715z
19
1725z
1725z
20
1730z
21
1735z
22
1745z
23
1750z
24
1755z
25
1800z
26
1805z
27
1810z
28
1815z P A U S E
29
2045z March 22, 2005 Part II (2 ½ hrs later)
30
2050z
31
2055z
32
2100z
33
2105z
34
2110z
35
2115z
36
2120z
37
2125z
38
2130z
39
99(90)
2135z
40
2140z
41
2145z P A U S E
42
1725z June 27, 2005
43
1730z
44
1735z
45
1740z
46
1745z
47
1750z
48
1755z
49
1800z
50
1805z
51
1810z
52
1815z
53
1820z
54
1825z P A U S E
55
How Can We Utilize This?

• Develop a WSR-88D algorithm that shows CG
  lightning/ numerous strike probabilities for
  detected cells.
• Currently, SCAN only has detected CG rate.
• Have algorithm update after every slice.
• Develop a web page display of CG lightning/
  numerous strike probabilities, based on WSR-88D
  data.
• Private sector could develop program to alert
  customers based on desired probability level.
• Issue CG lightning alerts for communities.
• Some offices include lightning potential
  (especially excessive strikes) in significant
  weather alert SPS products.

CG Lgtng Any Nmrs 95 70
75 20 40 5
L4 C8 J3
CG Lightning Probability

• 20 70 20 90
• 20 40 70
• 50 80 75
• 20 25
• 50 75
• 85
• 95
• 20
• 90 60

56
THE NEXT STEP
Develop algorithm, that produces radar-based CG
lightning (and numerous strike) probabilities,
for field evaluation.
QUESTIONS / COMMENTS ???

• Pete Wolf, SOO peter.wolf_at_noaa.gov
• NWS Jacksonville FL 904-741-5186
  ext 224