95th American Meteorological Society Annual Meeting Phoenix, AZ
6 January 2015
Slide 2
What are the Petals? How Are They Used Who Uses Them Example
Case Verification
Slide 3
Observed CG Lightning strikes SCIT history (movement and
intensity) Convective indices (helicity) Gamma distribution
function (alpha,beta) Cosine function ( f ) Formulated together in
GFE (Graphical Forecast Editor) to create a 2 hour thunder
probability grid (images at the top are actual rose petal images
which are very symmetric. Plumes are typically not due to wind
shear. We currently make Petals)
Slide 4
Cosine function Provides uniform distribution on either side of
the center line of the cell motion track. Uses SCIT cell movement
history to determine width of the petal (longer history =
confidence = more narrow petal) Gamma distribution function
Provides decreasing probabilities downstream. Alpha and Beta
parameters determine the rate of the decrease and are adjusted
based on CG rate and Cell speed.
Slide 5
In this Petal there are three Observed CG strikes. They are
shown as the red grids in the lower left corner. Thunder
probabilities are set to 100 at those grids. Storm motion is from
250 (Southwest). The yellow contours are the downstream Thunder
probabilities
Slide 6
Same Petal as on previous slide except this is filled to show
the thunder probability gradients. Dark purple is 100%, yellow is
around 25%.
Slide 7
Timing Balloons/Petals
Slide 8
Update Probability of Severe (ProbSvr) and Tornado (ProbTor)
Warning Issuance (locally defined as the probability that a warning
will be issued for that grid) Conditional (given a Thunderstorm)
Probabilities for Severe (CondProbSvr) and Tornado(CondProbTor) and
the background Thunderstorm probability (ProbThunder) grids are
updated by forecasters
Slide 9
Petals update the 1 and 2 hour Thunderstorm probability grids
(ProbThunder1) As petals are updated every 10 minutes, the ProbSvr
and ProbTor grids are updated. ProbSvr=ProbThunder1 * CondProbSvr
ProbTor=ProbSvr * CondProbTor Used to update 1 hour PoP and Wx
grids
Slide 10
OkArkSkyWarn.org. EMs can base text/email alerts based on
ProbSvr values. County based maximum values currently with points
supported through (OHMARS) In Severe weather briefings during
ongoing events Enhanced short term 1 or 2 hour Wx and PoP grids
Public via DSP (chicklet) page http://www.srh.noaa.gov/tsa/?n=dsp
Private Companies (CommPower and iNotify) See Poster #3 Poster
session #1 Kenneth Galluppi on Improving Weather Messaging
Slide 11
Dry line moving from west to east across the area. Tornado
watch issued at 19z First storm begins to develop at 1930z just
southwest of the Tulsa CWA
Slide 12
A. Original ProbThunder B. Petals A v B = Max (A,B) Petal
Enhanced Probabilities Downstream From Obs CG
Slide 13
A= Conditional probability That a given thunderstorm Will
require a Severe Thunderstorm Warning B= ProbThunder grid from
previous slide A * B = Probability that A Severe Thunderstorm
Warning will be issued. Enhanced by the Petals at 1940Z on Apr 27,
2014
Slide 14
Original ProbSvr. Higher Probabilities in reds and lower in
yellows and greens. Point A is Just ahead of forecast dryline and
Point B is just behind it. 38 16
Slide 15
Petals Only (no background values) Red grids are observed CG
The contours range from 100% around the red CG obs points, to 20%
on the outside edge. Each Petal is a 2 Hour Forecast of the
probability of a Thunderstorm. 4 Hour loop from 1930 2220 UTC
Slide 16
Slide 17
ProbSvr loop From 1930-2310 Cells originally develop directly
on the dryline over Point B. A little west of the higher
probabilites. The petals quickly updated the probabilities SVR and
TOR warning polygons take the grid values to 100%. You can see
petal influence ahead of the warning issuance.
Slide 18
Point A ProbThunder and ProbSvr Time Series
Slide 19
Point B ProbThunder and ProbSvr Time Series
Slide 20
Verification Petals yellow contours Obs CG in reds/ pinks Obs
CG shaded by # strikes Red = 10+ Strikes are given a 10km radius
Thunderstorm is defined as hearing thunder. Stronger long lived
cells verify best Cell movement (SCIT) biggest source of error
Slide 21
Later in event cells became more organized and verification
improved (lower Brier) along the center of the petal where higher
probabilities were. Fewer overall grids in the petal verified but
those that didnt were mainly on the outer edges.
Slide 22
Lightning Derived Thunderstorm Petals Do Provide An Added
Service A Good First Effort to Create an Automated Method To Bridge
the Watch to Warning Gap With Decent Verification Numbers Enhances
First 2 Hours of the Forecast Where HRRR/RAP Are Not As Useful.
Shared with Pleasant Hill, MO (EAX), Charleston WV (RLX), Salt Lake
(SLC) and Others.
Slide 23
Dr. Vincent Dimiceli (Deceased 2013) Associate Professor of
Mathematics Oral Roberts University, Tulsa, OK