An Objective Tool for Identifying Hurricane Secondary Eyewall Formation

1
An Objective Tool for Identifying Hurricane
Secondary Eyewall Formation Jim Kossin and Matt
Sitkowski Cooperative Institute for
Meteorological Satellite Studies University of
Wisconsin Madison, WI kossin_at_ssec.wisc.edu
62nd Interdepartmental Hurricane Conference
Charleston, SC, March 2008
2
This work is supported by the National Oceanic
and Atmospheric Administration under the GOES-R
Risk Reduction program and the Office of Naval
Research under Grant No. N00014-07-1-0163
3

• Goal
• Create a tool that uses readily available data to
  estimate the probability of secondary eyewall
  formation events in tropical cyclones
• Motivation
• These events are generally associated with marked
  changes in the intensity and structure of the
  inner core
• rapid intensity deviations
• significant broadening of the surface wind field
• changes in storm surge, sea-state, radius of 50
  kt wind
• Despite the importance of secondary eyewall
  formation in tropical cyclone forecasting, there
  is presently no objective guidance to diagnose or
  forecast these events.

4
Data and Method

• Our first step was to utilize the SHIPS
  developmental data.
• ambient environmental features
• geostationary satellite-derived features

The features were then separated into 2 classes
(using microwave, radar, recon reports, anything
available)
1) a secondary eyewall formed at some time in the
following 12 h 2) a secondary eyewall did not
form at any time in the following 12 h
Classes were limited to Category 1 hurricanes or
greater, with centers over water. 10 years
(19972006).
5
The algorithm is based on the Bayes probabilistic
model

• P (Cyes F) estimates the probability of
  imminent secondary eyewall formation, given the
  set F of observed features.
• P (Cyes) is the climatological probability (10
  in the North Atlantic).

Based on class separation
6
Leave-one-season-out cross validated algorithm
performance
20 (25)
22 (15)
12 (7)
970 (984)
52 (45)
Inclusion of IR increases the confidence of the
model
skill
skill
7
Cross validation hindcast example (hits, misses,
false alarms)
8
13 Sep 06Z (too early)
Misses and false alarms?
Ivan, 13 Sep 2004, 06Z
12 h later ?
9
The algorithm has been alerting us to secondary
eyewall formation events that we had previously
missed
10
False alarm?
27 Oct 1998, 00Z
URNT12 KNHC 260508
DETAILED VORTEX DATA MESSAGE A.
26/0508Z
B. 16 DEG 20 MIN N
81 DEG 53 MIN W

C. 700 MB 2391 M
D. NA
E. NA

F. 064 DEG 124 KT
G. 342 DEG 9 NM
H. 922 MB

I. 14 C/ 3047 M
J. 19 C/ 3024 M
K. 15 C/ NA

L. CLOSED WALL
M. CO8-15
N. 16 DEG
20 MIN N
81 DEG 53 MIN W
26/0508Z
O.
12345/7
P. .25/2 NM
Q. AF966 1113A
MITCH OB 09 KNHC DETAILED
MAX FL WIND 124 KT NW QUAD 0500Z. GOOD RADAR
PRESENTATION. DOUBLE EYEWALL, WIND CENTER
3 NM DIA.
Mitch, 27 Oct 1998, 00Z
12 h later ?
11
A miss and a false alarm
12
False Alarm
13
Hurricane Isabel (2003)
12 Sep 00Z ? hit 17 Sep 00Z ? miss
14
22 Sep 00Z ? hit 23 Sep 06Z ? miss
Hurricane Rita (2005)
15

• Work in progress
• Extension beyond SHIPS features
• Application to EPAC WPAC tropical cyclones
• Apply results to numerical simulations of
  secondary eyewall formation to better understand
  the physical mechanisms at work

We hope to piggyback a beta-version onto SHIPS as
soon as possible
16
End
17
(No Transcript)