Hurricane Hazard Information for

1
Hurricane Hazard DataforCaribbean Coastal
ConstructionProject Overview
Funded by the Caribbean Regional Program of the
US Agency for International Development. Executed
by the Organization of American States, in
conjunction with the Engineering Institute of
the University of the West Indies. System Design
and Analysis by Watson Technical Consulting.
2
Introduction
Project Goals Storm Hazard Assessment
Methodology Statistical Methodology Data
Bases On-line Data Access System
3
Project Goals
To create an easy to use on-line resource to
allow planners and coastal engineers to access
basic hurricane hazard data in a format conducive
for use in the design process. Hazards reported
are wind, wave, and storm surge, for 10, 25, 50,
and 100 year return periods. Return period data
is characterized at maximum likelihood (MLE),
75, 90, and 95 projection limits. Two Web
Sites to be established Primary University of
the West Indies Backup Old Dominion University
4
Storm Impacts at the Coast
Three phenomena output Wind, Waves, and Storm
Surge. Storm surge outputs include of wave setup,
wind setup, pressure setup, and astronomical
tide.
Wind
Wave Run up
Still Water Level at Shoreline
Wave Crest/TWL
Wave Setup
Wind Setup
Pressure Setup
Astronomical Tide
Mean Low Water
5
Methodology Overview
Simulate every Atlantic storm since 1886 (1851)
with the TAOS hazard modeling system to generate
the wind, wave, and storm surge at every grid
cell in the study area. At each point, compute
weibull distribution for each phenomena. Simulate
return period events to test interactions. Compu
te desired return period and projection limit
results.
6
The Arbiter Of Storms (TAOS) hazard model
Modular, 4D composite model platform with user
selectable modules 12 Wind Models 3 Boundary
Layer Models 3 Storm Surge Models 5 Wave
Models 2 Rainfall/Runoff Models - 1,620 basic
combinations - with damage functions, over
10,000 possible outcomes for a single
event! Scalable, from single processor through
supercomputer MPI or PVM systems such as the WTCI
earthdome system. TSAP Statistics System with
plug in modules for various distributions,
either desktop of batch run capability.
7
Storm Hazard Model Modules

• Key TAOS/mpi Modules
• Wind Field
• Water Flow (Storm Surge)
• Waves
• Tides (UTex CSR Model)
• Rainfall and runoff

• Damage
• Structure Damage
• Coastal Erosion
• Vegetation Damage
• GIS/Mapping Interface
• GrADS Meteorology package
• ArcInfo and ArcView
• Mapinfo
• GRASS

Storm components are interactive and can have
feedbacks for example, changes to land cover
cause changes to wind speeds, etc.
8
Input Data Bases for Simulations

• Digital Elevation Model (topography and
  bathymetry) Allows proper flow of water to be
  computed for storm surge, waves, and inland
  flooding
• Land Cover/Land Use (land and underwater) Frictio
  n effects on both air and water to
  compute correct wind speeds and flow
  rates Debris generation
• Historical Storm Characteristics for Statistical
  Analysis and validation

9
Model Physics Wind
Module Used Standard Project Hurricane
(SPH) Source NOAA Technical Report NWS 23
Basic wind field characteristics Asymmetric
wind field Radius of Maximum Winds Storm
motion Central and Far Field Pressure Terrain
effects The wind at a point depends on what
is upwind of the point, as much as 10
miles. Note storm centered grid system.
10
Wind Model Validation Example
Correlation 0.9222
Comparison of observed and computed peak winds
for Hurricane Floyd (1999)
11
Model Physics Storm Surge
Modules, vertical coordinates (3D vs. 4D),
horizontal resolutions depends on underlying data
bases Low Res (gt 600 meter grids) vertically
integrated equations Source Equations
Harris High Res (lt 600 meter grids) 5 vertical
layers Source Equations Watson, based on
Mellor Tides University of Texas CSR Tide Model
12
Comparisons with Tide Gauges
St. Augustine
Trident Pier
Solid Line Modeled Dots Observed
13
Peak Surge Observations (Marilyn, 1995)
Observed peak water levels from post storm
surveys on St. Croix and St. Thomas by the USACE
14
Model Physics Waves
NOAA Wavewatch III Model (deep and transitional
water) Source Tolman, modularized and modified
for moving grid WTC inshore model (shallow
water) Sources Lyons, Watson Again, physics
used depends on resolution.
15
Wave Validation
Comparison of operational (real time) wave
forecast and NOAA buoy reports
16
Validation Summary
Tests against detailed data from over 30 storms
worldwide. Performance summary (90
limits) Peak Wind (350 observations) /- 5
knots Peak Wave (185 observations) /- 0.8
meters (deep water Hs) Peak Storm Surge (1500
observations) /- 0.3 meters Model uncertainty
is included in the projection limit calculations.
17
Statistical Terminology
MLE Maximum Likelihood Estimate. The best
guess at a value. This is NOT the same as the
50 estimate. Projection Limit For a given
projection limit, the value which should not be
exceeded more than 1-(limit) of
observations. Example 100 year, 75 projection
limit. For a large number of 100 year periods,
75 of them will fall at or below this value.
Only 25 should be greater. OR, for a large
number of sites over a 100 year period, 75 of
the sites will see less than this value 25
would be expected to exceed it.
18
Statistical Analysis
Once storm runs are completed, outputs are
analyzed to produce return period data
sets. Single Site Example Four Seasons
Hotel, Nevis
19
Single Site Analysis Wind
Raw text output of TSAP program
BASIC TC WIND STATISTICS REPORT FOR Nevis
1220 EVENTS EXAMINED FROM 1850 TO
1999 NOTES RAW DATA IS IN FILE
YRMAX.DAT CAT 0 TROPICAL STORM EVENTS BY
CATEGORY SAFFIR/SIMPSON CATEGORY 0
48 SAFFIR/SIMPSON CATEGORY 1 9 SAFFIR/SIMPSON
CATEGORY 2 4 SAFFIR/SIMPSON CATEGORY 3
1 SAFFIR/SIMPSON CATEGORY 4 1 SAFFIR/SIMPSON
CATEGORY 5 0 -------------------------------
----------------- EMPIRICAL ESTIMATE OF RETURN
PERIODS SAFFIR/SIMPSON CATEGORY 0 2.5 YEARS
0.396 SAFFIR/SIMPSON CATEGORY 1 8.3
YEARS 0.121 SAFFIR/SIMPSON CATEGORY 2
24.8 YEARS 0.040 SAFFIR/SIMPSON CATEGORY 3
74.5 YEARS 0.013 SAFFIR/SIMPSON CATEGORY
4 149.0 YEARS 0.007 SAFFIR/SIMPSON
CATEGORY 5 Infinit YEARS 0.000
2 PARAMETER WEIBULL ESTIMATE OF RETURN
TIMES ALPHA 1.333316, BETA 35.105717 SDA
0.21 SDB 4.40, CORREL 0.871427 CHI2
13.486692 K-S 0.046980, K-S PROB 0.995793
5 YEAR WIND 50KTS (EMP 52KTS) 10 YEAR WIND
65KTS (EMP 70KTS) 25 YEAR WIND 84KTS (EMP
79KTS) 50 YEAR WIND 97KTS (EMP 89KTS) 100
YEAR WIND 110KTS (EMP 91KTS) MLE BASED WIND
RETURN TIMES WIND SPD PROB RETURN PD
50KTS 0.2014 5.0 YRS 64KTS
0.1078 9.3 YRS 100KTS 0.0176
56.7 YRS ----------------------------------------
-------- PROJECTION LIMIT CONFIDENCE LEVELS
YEAR 50 75 90 95 99
10YR 66.4 69.0 71.4 73.2 76.8
25YR 85.3 89.3 93.5 96.9 108.8
50YR 98.5 104.1 110.8 116.7
132.7 100YR 110.7 118.9 128.7 136.8
157.3
20
Single Site Statistics Nevis Wind
Graphic output of weibull fit for winds at Four
Seasons, Nevis
21
Statistical Methodology Validation
Used data from 1886-1987 to make a 10 year
forecast. Validated wind forecast against 10,600
land sites in the Atlantic Hurricane basin using
the 10 year period 1988 - 1997. Percentage of
sites below prediction limit Predicted Observ
ed 50 53 75 75 90
90 95 94 99 97
22
The Caribbean Hazard Data Web Server
Interactive, web based system using open
source/public domain software.
Temporary URL http//weather.methaz.com/cdcm/ Pe
rmanent URL UWI ?? ODU http//cdcm.cee.odu.edu/
Primary site will provide for mailing lists to
discuss coastal engineering and design
problems, techniques, etc.
23
On-line Documentation

• Notes on wind, wave, and storm surge model
  outputs
• Validation and uncertainty
• Terminology
• Building Code Notes

24
Available Data Bases
Regional 30 arc second (nominal 926 meters) As
part of the Caribbean Disaster Mitigation Project
(CDMP), a regional storm risk atlas was
developed. The return period was based on an
analysis of hurricanes from 1886 to 1998, with
storm hazard model runs conducted using a
regional 30 arc second (nominal 926 meter) grid.
It covers the Eastern Caribbean from Puerto Rico
through Trinidad and Tobago. 6 arc second (182
meter) Kitts/Nevis/Antigua/Barbuda This data set
was developed for the Post Georges Disaster
Mitigation Project. It covers the islands of St.
Kitts, Nevis, Barbuda, and Antigua. The return
period analysis was based on data from 1851
through 1999. The storm hazard model runs were
made at a resolution of 6 arc seconds (182 meter
grid). Only MLE data available for this area.
Hurricane Track Data Base This is a version of
the 1851-2000 storm track data set created by
from the US National Hurricane Center.
25
Data Set Access

• User may select return period and projection
  limit
• Choice of background maps
• Point and click report generation

26
Report Format
Site Map and Data Location
Site information
Hazard Data
27
Summary

• On-line hurricane hazard data for the Eastern
  Caribbean
• 10, 25, 50, 100 year return period data bases
• MLE, 75, 90, and 95 projection limits for
  regional data set
• RD site available at http//weather.methaz.com/c
  dcm/
• Final data sets and user interface by end of
  September 2001
• Final web sites at UWI and ODU by end of 2001