Evolving to Polygon Warnings: An Effective False Alarm Solution for the National Weather Service

1
Evolving to Polygon WarningsAn Effective
False Alarm Solutionfor the National Weather
Service

• Pete Wolf, SOO
• NWS Jacksonville FL

2
False Alarms

• Two Key Types
• Unavoidable FAs
• FIRE!, with smoke pouring into the
  room.
• Our mission requires warnings, regardless of
  verification.
• Unnecessary FAs
• FIRE!, with no smoke or fire
  (no actual threat).
• This is what we aim to reduce with polygon
  warnings.
• False Alarms are a Factor.
• Affect credibility, thus response.

3
Options for Reducing False Alarms

• Option 1 Issue fewer warnings (GPRA FAR goal)
• Change warning methodology that leads to fewer
  warning issuances (e.g. only strong meso or TVS
  for tornado warnings).
• For 3000/4000 unverified warningsFAR 0.75
• For 2300/3300 unverified warningsFAR 0.70
  (GPRA met)
• For 1500/2500 unverified warningsFAR 0.60
• Risk of significant unwarned events increases.
• Harold Brooks presented POD-FAR relationship.
• A bad idea from risk management perspective.
• Option 2 Issue smaller warnings
• Can reduce false alarm area by 50 or more.
• Still issue warnings for perceived threats.
• From risk management perspective, this is a
• much better idea.

4
Polygon Warning Defined

• Simply put, a polygon warning (PW) is the the
  following
• PW WCW FAA (unnecessary)
• where WCW is the whole county warning and FAA is
  the
• area where no perceived imminent threat exists.

-
In the polygon approach, all were trying to do
is reduce alerted territory where no imminent
threat is perceived.
5
FAR vs FAA (Not the Same Thing!)

• FAs Average FA size Total FAA
• 1000 800 sq mi each 800,000 sq mi
• 1500 266 sq mi each 400,000 sq mi

If you increase FAs from 1000 to 1500, you
increase your FAR by 50. But, at the same time,
if you reduce average warning size from 800 sq mi
to 270 sq mi each (1/3 original size), you
actually decrease FAA by 50. In a national
county- vs polygon-based warning study, from 2004
to 2005, FAA decreased by about 10, despite a
year-to-year increase in FAR.
6
Comparing Counties to Polygons

• Field evaluation was completed in 2005.
• Involved over 20 WFOs across the country.
• Results confirmed those from local studies at
  Wichita and Jacksonville.
• 70 reduction in TOR FAA with polygon warnings
  compared to whole-county warnings.
• Larger in WR offices, smaller in ER offices.
• Reduction is unnecessary false alarm area.
• Some negative impact on POD.
• More so than anticipated based on concept.
• Training and software issues.

T T
7
Polygon Warning Issues

• Issue 1 Describing polygon aerial coverage
• for text products/audio broadcasts (NWR).
• Text products will look similar to today.
• Likely will shift focus from pathcasts to
  locations
• in warning in text product.
• Any issues due to polygons also exist today
• with partial county warnings.

A B G H C D I J E F K L
1 2
8
What do we provide today?
The National Weather Service in Jacksonville has
issued a Severe Thunderstorm Warning for the
following locations Southern Nassau County in
Northeast Florida Western Duval County in
Northeast Florida Northern Clay County in
Northeast Florida until 700 pm EST. At 610 pm
ESTDoppler Radar indicated a line of
severe thunderstorms extending from 6 miles
southweest of St. George to near Baldwin to 7
miles south of Lawteyor extending from 15 miles
west of Callahan to 18 miles west of Downtown
Jacksonville to 15 miles southwest of
Middleburgmoving east at 40 mph. The storms are
capable of producing damaging winds up to 60 mph,
heavy rainfall and deadly cloud-to-ground
lightning. The storms will be near Bryceville
at 620 pm EST Nassau Village and Jacksonville
Heights at 635 pm EST Jacksonville International
Airport at 650 pm EST Riverside at 655 pm
EST This is a dangerous line of stormswith a
history of producing damaging winds. Take cover
now!
9
Polygon Warning Issues

• Issue 2 How will we verify polygon warnings?
• Shift to polygon-based verification (including
  GPRA measures) planned by FY08.
• WFOs will be encouraged to shift from counties to
  polygons in FY07.
• Concern about Polygon-based FAR leading to larger
  polygons and more FAA (not less FAA). Polygon
  size needs to be incorporated.
• Report accuracy is critical to for polygon
  warning verification.
• Will increase post-event work.

TORNADO WARNING
10
Polygon Warning Issues

• Issue 3 Software
• Private Sector game of chicken who goes
  first?
• Technology companies may wait until NWS shifts to
  polygons before upgrading their technology
• NWS may be waiting to shift to polygons until
  private companies upgrade their technology.
• After the NWS takes the lead, the private sector
  will rapidly follow.
• Who wants to be the only one not providing
  polygon warnings?
• NWS
• WarnGen
• LSR
• Changes in support of polygon warnings are
  planned in OB8 (Spring/Summer 07)

11
Polygon Warning Issues

• Issue 4 Forecaster TrainingPolygonology
• Areas in gap between polygons are not warned.
• Overlap not as critical of a problem but should
  be minimized as much as possible.
• County borders are not important, but CWA borders
  are!
• Confusion from issuing too many warnings.
• Dealing with multiple threats.

What single change could you make today at your
WFO to get us started in the right direction?
12
Avoid Drawing Whole County Polygons

• Stop drawing polygons to match county borders!
• Let the polygons outline the threat areas.
• (Lets review an interesting sr-supercell
  caseMarch 12, 2006.)

13
Warning-By-Polygon Benefits

• Focus warning on the true threat area
• Reduce the area alerted thats not threatened
• Easier handling storms moving along county
  borders
• Supports weather radio evolution to
• alert areas smaller than county
• Supports efforts to warn on smaller
• events we may typically miss/ignore.
• Supports development of graphical/web
• based products that show threat area.

14
Probabilistic Guidance The Next Step Beyond
Polygon Warnings

• Case 1 Central Mississippi, April 6, 2003
• Major severe weather event during the
  afternoon/early evening hours.

Blue Polygon Low Probability (warning may be
needed if further intensification/advection
occurs) Magenta Polygon High Probability (determ
inistic warning polygon) (Note focus is just
on damaging weather, not tornado vs wind vs
hail.)
15
2127z
2127z
16
2132z
17
2137z
18
2142z
19
2147z
20
2152z
21
2157z
22
2202z
23
2207z
24
2212z
25
2217z
26
2222z
27
2227z
28
2232z
29
2237z
30
2242z
31
2247z
2247z
32
2252z
33
2257z
34
2302z
35
2307z
36
2312z
37
2322z
38
2327z
39
2332z
2332z
40
2337z
2337z
41
2342z
42
2347z
43
2352z
44
2357z
45
0002z
46
0007z
47
0012z
48
0017z
49
0022z
50
0027z
51
0032z
52
0037z
53
0042z
54
0047z
55
0052z
56
0057z
57
0102z
58

0107z
59
0112z
60
0117z P A U S E
61
Probabilistic Guidance The Next Step Beyond
Polygon Warnings

• Case 2 Northern Florida, June 16, 2005
• Strong (isolated severe) pulse t-storms during
  the afternoon.

Green Polygon Low Probability (warning may be
needed if further intensification/advection
occurs) Magenta Polygon High Probability (determ
inistic warning polygon)
62
1734z
63
1744z
64
1754z
65
1804z
66
1814z
67
1824z
68
1834z
69
1845z
SVR issued for Suwannee Co
70
1855z
SVR issued for Suwannee Co
71
1905z
SVR issued for Suwannee Co
72
1915z
73
1925z
74
1935z
75
1945z
76
1955z
77
2005z
78
2015z
79
2025z
SVR issued for St. Johns County
80
2035z
SVR issued for St. Johns County
81
2045z
SVR issued for St. Johns County
82
2055z
SVR issued for St. Johns County
83
2105z
SVR issued for St. Johns County
84
2115z
SVR issued for St. Johns County
85
2125z
SVR issued for St. Johns County P A U S E
86
Latest Polygon Warning Implementation Information

• Polygon Evaluation Team
  Polygon Implementation Team
• Approved by Ops. Committee of Corporate Board
• Forecaster Training
  Planned through CY07
• Intro concept module/teletraining
• WES-based scenario training
• Needed software improvements Planned
  starting in OB8 (CY07)
• Operational implementation Planned for FY08
• Implementation of polygon Planned for FY08
• verification in GPRA

87
Final Thought

• One doesnt discover new lands without
• consenting to lose sight of the shore for
• a very long time. --- Andre
  Gide

Any thoughts/questions? Pete Wolf, SOO
NWS Jacksonville FL