The Atlantic Hurricane Database Re-analysis Project

1
The Atlantic Hurricane Database Re-analysis
Project Results for 1851 to 1910 and 1992s
Hurricane Andrew

• Chris Landsea
• NOAA/Hurricane Research Division
• Miami, Florida, USA
• CSU/CIRA
• April 2nd, 2004

Co-Authors Craig Anderson, Noel Charles, Gil
Clark, Peter Dodge, Jason Dunion, Jose
Fernandez-Partagas, James Franklin, Paul
Hungerford, Charlie Neumann, Mark Zimmer, and the
NHC Best Track Change Committee Acknowledgments
Sim Aberson, Auguste Boissonnade, Emery Boose,
Mike Chenoweth, Hugh Cobb, Jose Colon, Neal
Dorst, Paul Hebert, Brian Jones, Lorne Ketch,
Cary Mock, Ramon Perez Suarez, David Roth, Al
Sandrik, and David Vallee.
2
The Atlantic Hurricane Database Re-analysis
Project

• What is the historical hurricane database and how
  is it utilized?
• Why revise HURDAT and how is it done?
• Databases provided
• Results thus far
• Example The 1886 Indianola Hurricane
• 1851 to 1910
• Hurricane Andrew

Acknowledgment NOAA grant (NA7P0369), a grant
from the Risk Prediction Initiative and
supplemental funding from the Insurance Friends
of NHC
3
NOAA Dork Logo
Track and Intensity Forecast Development and Ver
ification
4
Climate Variability and Change Studies
5
Building Codes and Insurance Rates
6
HURDAT - Developed in the late 1960s in support
of the Apollo space program
7
Hurricane Bret (1999)
8
Intensity Estimates and Best Track Values
9
Center Fix and Best Track
10
Smoothed Representation in Best Track
Jarvinen et al. (1984)
11
Observational Platforms for Atlantic Hurricanes
Neumann et al. (1999)

12
Incorrect intensity and location at landfall
13
Too Rapid During Last 6 Hours
14
Pressure-Wind Relationship
15
Atlantic Major Hurricanes
Bias-removed
Landsea (1993)
16
GPS Dropsondes
Measures the wind around and in
hurricanes from the aircraft to the oceans
surface
Mean Wind Profile
Wind in Hurricane Georges
Franklin and Black (1999)
17
Work of Jose Partagas Historical
Reconstruction from 1851-1910
18
Ship Tracks
19
Using Ship Observations to Estimate
the Hurricanes Location
20
Estimating Intensity with and without Aircraft
Reconnaissance
21
The Beaufort Wind Scale Beaufort Knots
Description Number 0 lt 1 Calm 1
1-3 Light air 2 4-6 Light
breeze 3 7-10 Gentle breeze 4
11-16 Moderate breeze 5 17-21
Fresh breeze 6 22-27 Strong breeze
7 28-33 Near gale 8 34-40
Gale 9 41-47 Strong gale 10
48-55 Storm 11 56-63 Violent storm
12 gt 63 Hurricane
22
ATLANTIC WIND/PRESSURE RELATIONSHIPS
Winds (knots) computed from regional P/W
relationships
1)For GLFMEX Wind(knots)10.627(1013-p)0.5640
Sample size 664 r0.991 2)For lt25N
Wind(knots)12.016(1013-p)0.5337 Sample size
1033 r0.994 3)For 25-35N Wind(knots)14.172(1
013-p)0.4778 Sample size 922 r0.996 4)For
35-45N Wind(knots)16.086(1013-p)0.4333
Sample size 492 r0.974 5)For Kraft
Wind(knots)14.000(1013-p)0.5000 Sample size
13 r ??
P(MB) GLFMEX lt25N 25-35N 35-45N
KRAFT P(MB) P(IN)
960 100 100 94 90
102 960 28.35
Estimating the Central Pressure (Pr Pc)/(Pe
Pc) exp(-RMW/R)
23
Storm Surge
24
Surveys of Wind-caused Damage
25
(No Transcript)
26
Metadata File for Storm 5, 1886
Estimated central pressure from Ho (1989) of 915
mb at landfall in Texas is incorrect since it
relied upon a pressure measurement from San
Antonio, Texas, which has been found to be a
surface pressure rather than a sea level pressure
value. Additionally, it is likely that this was
not a central pressure measurement either as
strong winds were still observed at the time of
lowest pressure and that winds only shifted from
northeast to southeast. This sea level pressure
measurement of 971 mb (corrected from the 948 mb
surface pressure value) at 19Z on the 20th
implies a central pressure of around 965 mb,
assuming that the RMW estimate of Ho (of 12 nmi)
is slightly too small (15 nmi utilized instead).
965 mb suggests winds of 94 kt from the Gulf of
Mexico wind-pressure relationship for a marine
exposure - 85 kt utilized in best track for this
inland location. Using methodology in Ho et al.
(1987) as modified by B. Jarvinen (personal
communication), a 6.5 hr transit time from
landfall to a position near San Antonio, and the
965 mb central pressure near San Antonio, a new
value of 925 mb at landfall is estimated for this
hurricane. This suggests winds of 133 kt from
the Gulf of Mexico wind-pressure relationship.
The estimate of 15 nmi for RMW is slightly
smaller than climatology (18 nmi) for this
latitude and central pressure (Vickery et al.
2000), supporting slightly stronger winds for
this particular storm. 135 kt is chosen as the
maximum sustained winds at landfall. This is
consistent with the very high storm tide and
extreme destruction in Indianola, Texas. Storm
surge modeling efforts with the SLOSH model (B.
Jarvinen, personal communication) indicate that a
925 mb central pressure and RMW of 15 nmi
provides reasonable matches to observed surge
values. Positions are altered slightly after
landfall to better account for passage of the
hurricane's center near San Antonio at 19Z on the
20th. A storm tide of 15' was reported for
Indianola, Texas in Roth (1997b).
27
Center Fix Data for 1886/5
1886/05 (Synoptic/intensity) Date
Time Wind/Dir Pressure
Location Source 8/14/1886 ???? UTC
80 kt/ESE-SW ???? mb 17.7N 67.5W "Gertie
M. Rickerson" 8/15/1886 0700 UTC ? kt/N-SE
1000 mb 18.8N 70.0W Santo Domingo
8/17/1881 2000 UTC 50 kt/NE-WS ???? mb
22.5N 80.0W Cardenas 8/18/1880 ???? UTC
50 kt/NE-SE 1005 mb 24.8N 82.0W Key
West 8/18/1886 0900 UTC 70 kt/NE-S
994 mb 24.0N 82.5W "Manhattan" 8/19/1886
0700 UTC 60 kt/NE-SE ???? mb 24.8N
83.0W "Arbutus" 8/20/1886 1400 UTC 70 kt/E
999 mb 28.4N 96.8W
Indianola 8/20/1886 1930 UTC ? kt/??
971 mb 29.6N 98.5W San
Antonio 1886/05 (Center positions) Date
Time Location Source 8/20/1886
1930 UTC 29.6N 98.5W San Antonio
28
Revised Best Track (HURDAT) Data for 1886/05
Month Day Hour Lat. Long. Dir.
--------Speed--------- ---------Wind-------
---- Pres. ------------Type-----------
August 19 0 UTC 25.6N 87.9W 280 deg 16
kt 18 mph 29 kph 80 kt 95 mph 150 kph
-- mb Hurricane - Category 1 August 19 6
UTC 25.9N 89.7W 280 deg 16 kt 18 mph 29
kph 90 kt 105 mph 170 kph -- mb Hurricane
- Category 2 August 19 12 UTC 26.3N 91.9W
280 deg 20 kt 23 mph 37 kph 100 kt 115 mph
180 kph -- mb Major Hurricane - Category
3 August 19 18 UTC 26.5N 93.3W 280 deg
12 kt 13 mph 22 kph 110 kt 130 mph 210 kph
-- mb Major Hurricane - Category 3 August 20
0 UTC 26.8N 94.5W 285 deg 11 kt 12 mph
20 kph 120 kt 140 mph 230 kph -- mb Major
Hurricane - Category 4 August 20 6 UTC
27.4N 95.6W 300 deg 11 kt 12 mph 20 kph
130 kt 150 mph 250 kph -- mb Major Hurricane
- Category 4 August 20 12 UTC 28.0N 96.6W
305 deg 10 kt 11 mph 18 kph 135 kt 155 mph
260 kph 925 mb Major Hurricane - Category
4 August 20 18 UTC 28.8N 97.6W 310 deg
11 kt 12 mph 20 kph 85 kt 100 mph 160 kph
965 mb Hurricane - Category 1 August 21 0
UTC 29.7N 98.4W 320 deg 11 kt 12 mph 20
kph 60 kt 70 mph 110 kph -- mb Tropical
Storm August 21 6 UTC 30.8N 99.3W 325
deg 13 kt 14 mph 24 kph 50 kt 60 mph 90
kph -- mb Tropical Storm August 21 12
UTC 32.0N 100.0W 335 deg 13 kt 14 mph 24 kph
35 kt 40 mph 60 kph -- mb Tropical
Storm August 21 18 UTC 33.5N 101.3W 325 deg
18 kt 20 mph 33 kph 25 kt 30 mph 50 kph
-- mb Tropical Depression
29
U.S. Landfalling Hurricanes /Date
Time Lat Lon Max SS RMW Storm
Central States


Wind Surge Pressure
Affected
5-8/20/1886 1200Z 28.0N 96.6W 135kt 4
15nmi --- 925mb BTX4
30
Minutes of the NHC best-track change commmittee
meeting Members Jack Beven, Jim Gross, Richard
Pasch, Ed Rappaport, and chair (Colin McAdie)
present. (meeting 5 22 March, 2002) The
committee met to continue consideration of the
most recent set of best-track changes provided by
Chris Landsea et al. The committee resumed
discussion with storm 1 of 1897, as agreed,
completed discussion on the second third
(1890-1899) of the material and then proceeded
through storm 5 (becoming 6) of 1901. Jack Beven
noted that the use of 3- or 4-cup anemometers in
the early portion of the record could affect wind
estimates. Although there may be a systematic
way to handle this, it is not clear whether this
has been accounted for. lt Yes, this has been
accounted for as best possible. In use during
the period being revised - 1851 to 1910,
though anemometers were mainly available from the
1870s onward at coastal stations - was the four
cup Robinson anemometer. However, the primary
difficulties with this instrument were its
calibration and its mechanical failure in
high wind conditions. Even as late as 1890, the
highest wind that could be reliably
calibrated with this instrument was only about 30
kt (from a whirling machine), due to lack of a
strict comparison with a known quantity of
stronger winds (Fergusson and Covert 1924). gt
31
Estimated average position and intensity errors
in best track for the years 1851-1910. Negative
bias errors indicate an underestimation of the
true intensity. __________________________________
_______________________________________   Situatio
n Dates Position Intensity Error Intensity
Error Error (absolute) (bias) _______________
__________________________________________________
________ Open ocean 1851-1885 120 nmi/220 km 25
kt/13 m s-1 -15 kt/-8 m s-1 1886-1910 100
nmi/185 km 20 kt/10 m s-1 -10 kt/-5 m
s-1 Landfall in 1851-1885 120 nmi/220 km 25
kt/13 m s-1 -15 kt/-8 m s-1 unsettled
area 1886-1910 100 nmi/185 km 20 kt/10 m s-1 -10
kt/-5 m s-1   Landfall at 1851-1885 60 nmi/110
km 15 kt/8 m s-1 0 kt/0 m s-1 settled
area 1886-1910 60 nmi/110 km 12 kt/6 m s-1 0 kt/0
m s-1 ____________________________________________
________________________________
32
Strongest U.S. Hurricane 1851 to 1910
33
Most Intense (Category 3, 4, 5) Continental
United States Hurricanes 1851-1910 /Date Time
Lat Lon Max Saffir- Storm Central States
Winds Simpson Surge Pressure Affected
5-8/20/1886 1300Z 28.1N 96.8W 135kt 4 15'
925mb BTX4 1-8/10/1856 1800Z 29.2N 91.1W
130kt 4 11-12' 934mb LA4 1-9/9/1900 0140Z
29.1N 95.1W 125kt 4 20' 936mb CTX4
10-10/2/1893 0800Z 29.3N 89.8W 115kt 4 ---
948mb LA4 7-10/2/1898 1600Z 30.9N 81.4W
115kt 4 16' 938mb GA4,DFL2 6-9/16/1855
0300Z 29.2N 89.5W 110kt 3 10-15' (950mb)
LA3,MS3 1-8/11/1860 2000Z 29.2N 90.0W 110kt
3 12' (950mb) LA3,MS3,AL2 4-9/1/1879 1600Z
29.5N 91.4W 110kt 3 --- (950mb) LA3
2-8/13/1880 0100Z 25.8N 97.0W 110kt 3 ---
931mb ATX3 3-8/16/1888 1900Z 25.8N 80.1W
110kt 3 14' (945mb) CFL3,BFL1 4-9/29/1896
1100Z 29.2N 83.1W 110kt 3 --- 960mb
AFL3,DFL3,GA2,SC1,NC1,VA1 10-10/12/1886 2200Z
29.8N 93.5W 105kt 3 12' (955mb) LA3,CTX2
9-10/13/1893 1300Z 33.0N 79.5W 105kt 3 14'
955mb SC3,NC2,VA1 5-10/9/1894 0300Z 30.2N
85.5W 105kt 3 --- (955mb) AFL3,GA1
3-8/18/1899 0100Z 35.2N 75.8W 105kt 3 ---
(945mb) NC3 8-10/18/1906 0930Z 24.7N 81.1W
105kt 3 --- 953mb BFL3,CFL3 8-9/21/1909
0000Z 29.5N 91.3W 105kt 3 15' 952mb LA3,MS2
4-8/23/1851 2100Z 30.1N 85.7W 100kt 3 12'
(960mb) AFL3,GA1 1-8/26/1852 0600Z 30.2N
88.6W 100kt 3 12' 961mb AL3,MS3,LA2,AFL1
2-9/8/1854 2000Z 31.7N 81.1W 100kt 3 ---
950mb GA3,SC2,DFL1 6-9/8/1869 2200Z 41.4N
71.7W 100kt 3 8' 965mb RI3,MA3,CT1
3-8/17/1871 0200Z 27.1N 80.2W 100kt 3 ---
955mb CFL3,DFL1,AFL1 5-10/7/1873 0100Z 26.5N
82.2W 100kt 3 14' 959mb BFL3,CFL2,DFL1
3-9/16/1875 2100Z 27.7N 97.2W 100kt 3 15'
(960mb) BTX3,ATX2 4-10/3/1877 0500Z 30.0N
85.5W 100kt 3 12' (960mb) AFL3,GA1
2-8/18/1879 1200Z 34.7N 76.7W 100kt 3 7'
971mb NC3,VA2 2-9/10/1882 0200Z 30.4N 86.8W
100kt 3 --- 949mb AFL3,AL1 2-8/25/1885
0900Z 32.2N 80.7W 100kt 3 --- (953mb)
SC3,NC2,GA1,DFL1 6-8/28/1893 0500Z 31.7N 81.1W
100kt 3 9-10' 954mb GA3,SC3,NC1,DFL1
4-7/21/1909 1650Z 28.9N 95.3W 100kt 3 10'
959mb CTX3 10-10/11/1909 1800Z 24.7N 81.0W
100kt 3 --- 957mb BFL3,CFL3
34
Longest lived hurricane on record
35
Four Hurricanes Simultaneously
36
U.S. Deadliest Hurricanes of the 19th Century
37
(No Transcript)
38
Busiest U.S. Hurricane Season on Record
39
Changes to U.S. Hurricanes - 1899 to 1910 /Date
Original 2003 HURDAT Category/State HURDAT
Revision Changes 2-8/1/1899 TS AFL2
Upgrade by two categories 3-8/18/1899 NC3 NC3
No change 8-10/31/1899 NC3,SC3 NC2,SC2
Downgrade by a category 1-9/9/1900 CTX4 CTX4
No change 3-7/11/1901 NC1 NC1 No change
4-8/14-15/1901 LA2,MS2 LA1,MS1,AL1 Downgrade by
a category, add in Alabama 1902 - No U.S.
hurricanes 3-9/1113/1903 CFL2,AFL1 CFL1,AFL1
Downgrade by a category in SE Florida
4-9/16/1903 NJ1,NY1 NJ1,DE1 No change in
category, add in Delaware, remove New York
2-9/14/1904 SC1 SC1 No change 3-10/17/1904
TS CFL1 Upgrade by a category 1905 - No U.S.
hurricanes 2-6/17/1906 CFL1 BFL1,CFL1 No
change in category, add in SE Florida
5-9/17/1906 SC3,NC3 SC1,NC1 Downgrade by two
categories 6-9/27/1906 MS3,AL3
MS2,AL2,AFL2,LA1 Downgrade by a category, add in
NW Florida and Louisana 8-10/18/1906 CFL2
BFL3,CFL3 Upgrade by a category, add in SW
Florida 1907 - No U.S. hurricanes 3-7/31/1908
NC1 NC1 No change 2-6/29/1909 TS ATX2
Upgrade by two categories 4-7/21/1909 CTX3
CTX3 No change 6-8/27/1909 ATX2 ATX1
Downgrade by a category 8-9/21/1909 LA4
LA3,MS2 Downgrade by a category, add in
Mississippi 10-10/11/1909 CFL3 BFL3,CFL3 No
change in category, add in SW Florid 3-9/14/1910
ATX2 ATX2 No change 5-10/17/1910 BFL3 BFL2
Downgrade by a category Notes Overall 2
hurricanes upgraded 2 categories, 2 hurricanes
upgraded 1 category, 10 hurricanes no change in
category, 7 hurricanes downgraded 1 category, 1
hurricane downgraded 2 categories.
40
Incomplete Records
41
Settled Area In 1850
42
(No Transcript)
43
Busy Atlantic Seaboard Decade The 1890s
44
Sandrik and Landsea (2003)
45
Three Major Hurricanes In Georgia
46
(No Transcript)
47
(No Transcript)
48
    Atlantic Hurricane Re-Analysis Project
             
http//www.aoml.noaa.gov/hrd/data_sub/re_anal.html

• Re-analysis results for 1851 to 1910
• Overview and Highlights
• Documentation
• Data
• 2. Hurricane Andrew's Upgrade
• 3. Books by Jose' Fernandez Partagas

1911 through 1940 being re-analyzed currently
(working with William Bredemeyer, Steve Feuer,
John Gamache and Lenworth Woolcock)
Picture from "Florida's Hurricane History", by
Jay Barnes
49
NOAA Dork Logo
50
(No Transcript)
51
But nobody lives at 10,000 ft. How can we use
flight-level data to estimate surface winds?
90 kt
95 kt
90 kt
105 kt
RECON FLIGHT-LEVEL WINDS HURRICANE GEORGES
9/20/98 20-23Z
52
In the early 1990s, reduction factors used by
NHC ranged from 75-90 of the flight-level wind.
Powell and Black (1990) concluded 63-73 for
700mb to surface reduction factor, but had few
eyewall high-wind cases.
53
EYEWALL SCHEMATIC
EYEWALL
EYE
AIRCRAFT TRACK
DROPSONDE TRAJECTORY
80
70
100
90
10000 ft
1-2 miles
54
AVERAGE OF 357 GPS DROPSONDE PROFILES IN THE
HURRICANE EYEWALL. ON AVERAGE, THE SURFACE WIND
IS 90 OF THE WIND AT 700 MB AIRCRAFT
RECONNAISSANCE LEVELS (75-80 AT LOWER ALTITUDES).
Franklin et al. (2002)
55
New HWIND Analysis For Hurricane Andrew 150
kt 93 of flight level
56
Some New Hurricane Andrew Data Radar Feature
Tracking
57
The Re-analysis of Hurricane Andrew (1992)

• Flight-level wind to surface extrapolation 145 kt
  (10)
• Feature tracking from the Miami radar 145 kt
  (15)
• Pressure-wind relationships 145 kt (20)
• Satellite intensity estimates 145 kt (20)
• Storm surge and SLOSH implications 145 kt (25)
• Structural damage survey estimates 135 kt (30)

58
Revised Best Track
59
A RE-ANALYSIS OF ANDREWS INTENSITY 1992s
Hurricane Andrew originally was assessed as a
Category 4 hurricane, based primarily upon
extrapolation of flight-level winds to the
surface. Research using GPS dropwindsondes in
the late 1990s and early 2000s has demonstrated
that stronger winds exist at the surface the
hurricane eyewall than originally believed. A
re-analysis indicates that Hurricane Andrew
reached Category 5 status at its landfall in both
the northern Bahamas and in southeastern Florida
(145 kt 10 kt).    Because of this
reclassification, the return period of
catastrophic hurricanes like Andrew increases
from about 30 years to around a century for
southeastern Florida. Thus the risk from
Andrew-like hurricanes there is significantly
LESS than previously estimated.
60
The Atlantic Hurricane Database Re-analysis
Project

• What is the historical hurricane database and how
  is it utilized?
• Why revise HURDAT and how is it done?
• Databases provided
• Results thus far
• Example The 1886 Indianola Hurricane
• 1851 to 1910
• Hurricane Andrew

Acknowledgment NOAA grant (NA7P0369), a grant
from the Risk Prediction Initiative and
supplemental funding from the Insurance Friends
of NHC
61
Settled Area In 1890
62
(No Transcript)
63
Atlantic Multidecadal Mode
Local Correlation of SST versus REOF
Mestas-Nunez and Enfield (1999)
64
(No Transcript)
65
WHAT IS THE INTENSITY OF A TROPICAL CYCLONE AND
HOW IS IT MEASURED TODAY?

• Maximum sustained surface wind Maximum wind,
  averaged over 1 minute interval at an altitude of
  33 ft (10 m), associated with the circulation of
  the tropical cyclone at a given point in time.
• With very, very few exceptions, direct
  observations of the maximum sustained surface
  wind in a tropical cyclone are not available.

• Satellite imagery using the Dvorak technique
• Aircraft reconnaissance flight-level winds
• GPS dropwindsondes

66
EYEWALL STRUCTURE CAN VARY SIGNIFICANTLY FROM
STORM TO STORM, OR EVEN DURING DIFFERENT STAGES
OF THE LIFE CYCLE OF A SINGLE STORM.
Franklin et al. (2002)
67
Step 1 Flight Level to a Mean Boundary Layer (0
to 500m)
MBL to 700mb Ratio
Distance as a Function of RMW
Dunion, Landsea, Houston and Powell (2002)
Dunion and Powell (2002)
68
Step 2 Mean Boundary Layer (0 to 500m) to the
Surface
Dunion, Landsea, Houston (2002) Dunion and
Powell (2002)
69
(No Transcript)
70
Hurricane Andrew - Satellite Dvorak Estimates 127
kt/935 mb
71
(No Transcript)
72
Hurricane Andrews Storm Surge And SLOSH Runs
73
Structural Damage Surveys of Hurricane Andrew
74
(No Transcript)