Lightning Forecasting

1
Lightning Forecasting at the Storm Prediction
Center
Joe Schaefer with help from Phillip Bothwell,
David Bright and Russell Schneider
2
Current uses of the NLDN by the SPC

• First strike detection convective initiation
• Subjective measure of convective intensity
  trend
• Verification and validation of thunderstorms
• Detection of convection in regions of sparse
  radar coverage (esp. for Fire Weather)

3
Lightning (CG) Fire Starts
4
SPC Lightning Forecasts

• General Thunder in Convective Outlook
• Enhanced Thunder
• Presently experimental on Internet
• Operational Product next year
• Fire Weather Outlook

Two types of Internal Thunderstorm Guidance
5
Enhanced Thunder 19 July 2005 200000Z 201200Z
6
Ensemble Forecasts

• Current SPC SREF processing
• All operational computations on NOAA CCS
• SREF products specialized for the national
  mission of the SPC
• 16 member SREF (1 time-lagged Eta member)
• Include the -3 hr operational Eta in all
  calculations
• Add the 3 hr operational Eta to spaghetti plots

7
CLOUD PHYSICS THUNDER PARAMETER Percent of
members with appreciable CAPE in the 0oC to -20oC
layer
SREF MEAN CPTP 1 (DASHED)
SREF PROB CPTP gt 1 (SOLID FILLED)
F15 SREF PROBABILITY CLOUD PHYSICS THUNDER
PARAMETER gt 1
8
- Pr (CPTP) gt 1 x Pr (PCPN) gt .01
Uncalibrated probability of lightning
F15 SREF 3-hr COMBINED PROBABILITY OF LIGHTNING
9
F15 SREF 3-hr COMBINED PROBABILITY OF LIGHTNING
(Calibrated) LIGHTNING STRIKES
10

• Statistical Forecasts
• Based on Phillip Bothwells Dissertation work
• Perfect Prog Forecast Technique
• Principal Component Analysis determines
  predictors
• Logistic regression
• 40 km grid resolution at 3 hour time intervals
• Uses Lightning Climatology (no knowledge of
  ongoing activity
• Forecasts
• Probability of 1 or more CG flashes
• Probability of 100 or more CG flashes.

11

• OBJECTIVES
• Technique applicable to ANY analysis or gridded
  data set
• (i.e., forecast for the next 3 hours 57 to 60
  hours,
• or any time period in between)
• System can run on data fromANY forecast model.
• inter-model comparisons (different modelse.g.,
  RUC and ETA)
• intra-model comparisons (different cycles of
  same model).
• Distinguish low lightning producers from high
  producers

12
12-15UTC
15-18UTC
LTG CLIMATOLOGY Probability of one or more CG
flashes / 40x40 km grid box / 3 hrs. Centered on
July 22
Relative minimum during daytime hours
Relative maximum during daytime hours
18-21 UTC
21-00UTC
13
00-03UTC
03-06UTC
LTG CLIMATOLOGY Probability of one or more CG
flashes / 40x40 km grid box / 3 hrs. Centered on
July 22
Relative maximum during overnight hours
Relative minimum during overnight hours
06-09UTC
09-12UTC
14
(No Transcript)
15
57-60 Hour Lightning Forecast 12Z July 13, 2004
Forecast
Observed
16
There can be a big difference between storms that
produce only one flash and
17
those that have large numbers of flashes!
Previous attempts at predicting events with
significant flash rates have not met with much
success!
NOAA Photo Library NWS Historical Collection
18
Significant CG Lightning Events100 or more CG
flashes/40x40 km/3 hrs

• Kempf and Krider (2003) as well as others have
  found a strong connection between daily rain
  volumes and corresponding counts of CG lightning
• Intense areas of CG lightning can often be
  associated with strong to severe storms

19
Probability of 100 or more CG Flashes 9-12 hour
Forecast 18Z Sep. 18, 2004
Climatology 03 Z 06Z 5 Days Centered on Sep. 20
Solid Probability 21,, 2, 5 Dashed Observed
100,200,399 Flashes
20
Future use of Lightning Data

• Become more quantitative
• Track development evolution of total lightning
  within individual thunderstorms
• Prediction of critical lightning characteristics
• Positive CG Strokes Fire Weather (house fires?)
• Forecasts of severe lightning events
• We need to learn more

21
(No Transcript)