Conference PaperPDF Available

Implications of Hurricane - Sea Surface Temperature Relationship

Authors:

Abstract and Figures

This paper presents a study to assess the impact of possible future climate change on the joint hurricane wind and rain hazard along the northeast US coastline. A postulated climate change model (IPCC scenario) was considered, which suggested changes in sea surface temperature (SST) (i.e., the driving parameter in most modern hurricane models). Relationships between SST and hurricane genesis frequency, genesis location, and track propagation were incorporated into state-of-the-art hurricane simulation procedures. Results from the SST conditioned hurricane simulations indicate the wind and rain hazards for the northeast US are likely to increase in a warmed climate, while the overall number of landfalling events is likely to decrease.
Content may be subject to copyright.
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
Implications of Hurricane Sea Surface Temperature Relationship
Lauren Mudd
Staff Engineer, IntraRisk Division, Applied Research Associates, Raleigh, NC, USA
Chris Letchford
Professor and Department Head, Dept. of Civil and Environmental Engineering, Rensselaer
Polytechnic Institute, Troy, NY, USA
David Rosowsky
Provost, Senior Vice President, and Professor of Engineering, University of Vermont, Burlington, VT,
USA
ABSTRACT: This paper presents a study to assess the impact of possible future climate change on the
joint hurricane wind and rain hazard along the northeast US coastline. A postulated climate change
model (IPCC scenario) was considered, which suggested changes in sea surface temperature (SST)
(i.e., the driving parameter in most modern hurricane models). Relationships between SST and
hurricane genesis frequency, genesis location, and track propagation were incorporated into state-of-
the-art hurricane simulation procedures. Results from the SST conditioned hurricane simulations
indicate the wind and rain hazards for the northeast US are likely to increase in a warmed climate,
while the overall number of landfalling events is likely to decrease.
The IPCC Fifth Assessment Report (Pachauri,
2014) states warming of the climate system is
unequivocal, and continued emission of
greenhouse gases will cause further warming and
long-lasting changes in all components of the
climate system, increasing the likelihood of
severe, pervasive and irreversible impacts for
people and ecosystems. The consideration of
extreme environmental event hazards in the
presence of such warming would allow for a
better understanding of the risk to our existing
inventory of civil infrastructure. A more
thorough understanding of future risks in turn
will ensure that target safety and performance
levels are met when designing structures and
infrastructure systems in the future.
For US coastal regions, specifically along
the Atlantic Ocean and Gulf of Mexico, a
quantitative assessment of climate change impact
on hurricane hazard performance levels is
needed. The northeast US coast was selected as
the sample study region herein for several
reasons. First, the future climate scenario used in
this study projects the largest increases over
modern day values of sea surface temperature
(SST) to occur just off the northeast US coast
(see Section 2). The second motivation for
choosing the northeast US as the study region
comes from the current (already high)
vulnerability of many areas in the region, such as
New York City and Boston. In addition,
increases in both population and development
along coastal areas are only expected to increase
the vulnerability of this region.
1. FUTURE CLIMATE PROJECTION
Representative concentration pathway (RCP)
scenarios have been developed recently for
climate change projections for the IPCC Fifth
Assessment Report and future editions. The
RCPs are projections of radiative forcing, based
primarily on the forcing of greenhouse gases.
This study utilizes RCP 8.5 as the future climate
scenario. RCP 8.5 is a high forcing scenario,
with 8.5 W/m
2
total radiative forcing in the year
2100, representing a case in which no technology
1
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
or policies have been implemented to reduce
greenhouse gas emissions. In comparison, the
2005 radiative forcing level, according to the
IPCC fourth Assessment report, is 1.6 W/m
2
.
The difference between the current (2012) SST
and the future (2100) projected SST in August,
typically the most active hurricane month, is
shown in Figure 1. The largest SST increases
occur just off the coast of the study region (i.e.,
the northeast US/Canadian coast).
Figure 1: Projected SST change under climate
scenario RCP 8.5 from August 2012 to August 2100.
2. PREVIOUS METHODOLOGY
This study builds upon state-of-the-art
probabilistic hurricane simulation procedures.
Key components include modeling of hurricane
genesis frequency and location, gradient wind
field, track propagation, central pressure, central
pressure decay, and rainfall.
Due to the high variability of frequency of
hurricanes from year to year, as well as physical
limitations of past observing and reporting
capabilities, there is no consensus regarding the
completeness of the HURDAT database
(Holland and Webster (2007); Mann and
Emanuel (2006); Mann et al. (2007); Knutson et
al. (2008); Vecchi and Knutson (2008); Landsea
et al. (2010)). However, US landfalling hurricane
records have been shown to be accurate from
1900 onwards (Landsea (2000)). Therefore, only
historical events that made landfall in the United
States are used in the development of the
hurricane models herein.
The simulation of a hurricane from genesis
until it is no longer hurricane strength follows a
Monte Carlo simulation procedure. For each
iteration (i.e., year) of the simulation, an annual
hurricane genesis frequency is generated
according to a Poisson process. Each simulated
hurricane is first generated in the Atlantic basin
with initial parameters (i.e., initial location,
heading angle, translational velocity, and central
pressure) based directly on a randomly selected
historical hurricane contained in the HURDAT
database.
The hurricane then moves along a track
defined by an empirical tracking model. In
developing the tracking model, the Atlantic basin
was first divided into a 5º square grid. Using
historical data contained in the HURDAT
database, linear regression was employed to
determine the heading and translational velocity
in each grid cell as a function of latitude,
longitude, and previous values of heading and
translational velocity. Further detail on the
tracking model can be found in Mudd et al.
(2014), Vickery et al. (2000b).
At each subsequent 6-hour interval, the
central pressure, gradient wind field and rainfall
are obtained. The central pressure model is
identical to that presented in Vickery et al.
(2000a), which in turn is based upon the relative
intensity concept presented in Darling (1991).
The gradient wind field model is presented in
Georgiou (1985). The rainfall model employed
here is a Weibull construct, wherein the scale
and shape parameters are dependent upon the
gradient wind field, the sea surface temperature,
and the location of the hurricane eye.
3. ANALYSIS: TIME AND SST
The gradient wind field, central pressure, central
pressure decay, and rainfall models used herein
include a T
s
term. Therefore, possible changes in
hurricane behavior due to climatological effects
can be assessed, using SSTs from both the
current climate and from future projected climate
scenarios as input to those models. The genesis
and track propagation models, however, do not
include T
s
and therefore cannot be investigated in
2
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
this way. In this section, trends in genesis
frequency, genesis location, and track
propagation are explored. First by examining
whether a temporal trend exists in the historical
record, and second by examining any
relationship that exists with SST.
3.1. Genesis Frequency
Genesis of hurricanes was simulated according to
a Poisson arrival process. With the decision to
only utilize landfalling hurricane records, linear
regression was used to fit a trend to the 20-year
moving average landfalling hurricane frequency.
As can be seen in Figure 2, no trend is apparent
(versus a clear increasing trend when utilizing
the complete HURDAT database). The annual
occurrence rate, for both the current (2012) and
future (2100) climate scenarios, was found to be
approximately 2.9 hurricanes per year.
Figure 2: Projected SST change under climate
scenario RCP 8.5 from August 2012 to August 2100.
An alternative method of investigating
climate induced changes in hurricane frequencies
would be to identify some trend in annual
occurrence rates directly with SST, instead of
indirectly with time. In order to do this, an
average value of SST in the main development
region (MDR) is obtained for each year by
averaging the historical SST values from July to
October over the North Atlantic MDR. The
MDR in this study extends from 5ºN to 25ºN
between the African and Central American
coasts.
First the relationship between the number of
US landfalling hurricanes and Atlantic basin
hurricanes with MDR SST was examined using
regression analysis. To minimize uncertainty
arising from the completeness of the HURDAT
database, only data from the start of routine
aircraft reconnaissance measurements (1948) to
the present is utilized. US landfalling hurricanes
show a weak correlation at the 5% significance
level (R
2
= 0.0715) when related directly to
MDR SST. Relating all hurricanes generated in
the Atlantic basin to MDR SST reveals a
significant trend at the 5% significance level,
with a higher R
2
value of 0.2217. Furthermore, a
significant relationship was also noted between
Atlantic basin hurricanes and US landfalling
hurricanes at the 5% significance level, with an
R
2
value of 0.2994.
Acknowledging that a significant
relationship exists at the 5% significance level
between the number of Atlantic basin hurricanes
with MDR SST and with the number of US
landfalling hurricanes, a prediction scheme is
developed using MDR SST to first predict the
annual number of hurricanes generated in the
Atlantic basin and then to predict the number of
Atlantic basin hurricanes that will actually make
landfall in the US. This analysis builds upon the
previous work of Jewson et al. (2008). Using the
two statistically significant relationships does not
change the relationship between the number of
US landfalling hurricanes and MDR SST, but
results in a narrowing of the 95% confidence
bounds. The final relationship between US
landfalling hurricanes and MDR SST is shown in
Figure 3. The genesis prediction scheme uses
linear relationships between MDR SST, Atlantic
basin hurricanes, and US landfalling hurricanes
to determine the Poisson arrival rate (λ) used to
generate pseudo-random annual hurricane
Figure 3: Relationship between US landfalling
hurricanes vs. MDR SST. Historical data (blue), best-
fit and 95% confidence bounds (red).
3
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
occurrence rates within the simulation
procedures.
3.2. Genesis Location
Historical hurricane genesis locations were
analyzed to identify possible temporal trends.
Unlike the genesis frequency analysis, only US
landfalling events were considered in the
assessment of genesis location. The genesis
locations, obtained from the HURDAT dataset
for the period 1851-present, were first divided
into eight approximately 20-year periods, as well
as three approximately 50-year periods. The 20-
year period was chosen as it has been shown to
cover both the active and inactive phases of the
decadal oscillation cycle. The 50-year period was
chosen, somewhat subjectively, in light of the
limited amount of historical data occurring in
some 20-year periods.
Histograms were then created using a 1º
square spatial grid of the Atlantic basin, and
recording the number of occurrences in each grid
cell for the period of interest. In order to make a
meaningful comparison of genesis locations over
time, the resulting histograms were normalized
for each time period of interest. The distribution
of genesis location in the Gulf of Mexico
becomes more concentrated further north with
time, while genesis location in the MDR remains
relatively unchanged. The average MDR SST
during the periods 1851-1911, 1912-1961, and
1962-2012 was 27.22, 27.38, and 27.54ºC
respectively. No spatial or temporal trends were
found using the 20-year segmented data, likely
due to the relatively small number of US
landfalling hurricanes captured in each time
period.
Similar to the normalized histograms from
the temporal analysis, probability density
functions (PDFs) were created to assess the
relationship between genesis location of US
landfalling events and MDR SST. This
assessment builds upon analysis of Hall and
Jewson (2007) and Hall and Yonekura (2013),
where genesis location was modeled using
steady-state Gaussian kernel density estimation.
Hall and Jewson (2007) utilized all hurricane
events occurring in the Atlantic basin from 1950
to 2003 (i.e., high quality data after the advent of
routine aircraft reconnaissance) as the basis for
their analyses. Hall and Yonekura (2013) then
used the Hall and Jewson (2007) genesis model
to assess what effect genesis location in a
warmed climate would have on the number of
hurricane events that made landfall in the US.
The genesis location PDFs were obtained by
the summation of two-dimensional Gaussian
kernel density estimates (Hall and Jewson
(2007)) as in Eq. 1, which were conditioned on
yearly averaged MDR SST.
=
=
N
i
L
i
D
NL
f
1
2
2
2
exp
2
2
1
),(
π
χψ
(1)
where
),
(
χψ
f
= the PDF at location at latitude
ψ
and longitude
χ
, D = the distance between
location
)
,(
χψ
and the i
th
genesis site, and L =
bandwidth of the genesis location PDF. In order
to condition the PDF on MDR SST, the historical
genesis locations were first binned according to
the yearly averaged value of MDR SST. MDR
SST bins with a range of 0.5ºC, centered every
0.1ºC were employed. Once the PDFs
corresponding to each value of MDR SST were
obtained, regression analysis was used to
determine the probability density of genesis
location as a linear function of SST.
The bandwidth of the PDF was optimized
by maximization of the coefficient of
determination R
2
. When the value of L is too
small, the genesis PDF is concentrated around
many local maxima; when the value of L is too
large, trends in the genesis location are smoothed
out. The optimal value of L was found to be 125
km, with an R
2
value of 0.86. Figure 4 shows the
normalized histogram of historical genesis
locations and the normalized Gaussian kernel
genesis location PDF, which compare quite
favorably.
With confidence in the MDR conditional
genesis location PDF to replicate the historical
data, a future genesis location PDF was
conditioned on the value MDR SST under RCP
4
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
8.5 for the year 2050 and 2100. The normalized
PDF for the year 2100 is shown in Figure 4. The
increase in MDR SST for the year 2100 is
1.90ºC. In general, as MDR SST increases, the
distribution of genesis location in the Gulf of
Mexico appears to be concentrated further north,
just off of the Gulf Coast of the US. Along the
east coast of the US, genesis concentration
decreases with increasing MDR SST.
Interestingly, the distribution of genesis location
in the MDR displays only slight shifts.
Figure 4: (top) Normalized histogram of historical
genesis locations of US landfalling hurricanes (1851-
2012), and normalized Gaussian kernel genesis
location PDF with a bandwidth of 125 km for
(middle) 2012 and (bottom) 2100.
3.3. Track Propagation
Two approaches were used to identify a trend in
hurricane track with time. The first approach
involved visual inspection of the historical
hurricane tracks themselves, to identify any
overall trends between the time periods chosen.
From the historical tracks of hurricanes that
made landfall in the US, segmented into
approximately 50-year periods, no trend was
evident in the behavior of the hurricane tracks
with time. Next, the eastern US coast and Gulf of
Mexico were segmented into five different
regions, and landfalling rates in each region were
analyzed over time in order to determine if there
existed any obvious change in the spatial or
temporal distribution of landfall locations with
time. No trend in hurricane track over time was
seen for any area. Additionally, the spatial
distribution of landfalls remains relatively
constant over all time segments. Therefore, no
clear trend in hurricane track could be identified
qualitatively (e.g., changes in tracks over time)
or quantitatively (e.g., changes in landfall rates
over time).
Several studies (Elsner (2003); Elsner &
Jagger (2006); Hall & Yonekura (2013)) have
shown hurricane tracks to behave differently in
varying climate states. Having found no trend in
hurricane track propagation with time, this study
then analyzed the variation in hurricane track
behavior directly with SST. In order to do so,
three variations of the empirical tracking model
were considered. The first candidate model
included the effects of the relationship between
SST and translational velocity; the second
candidate model included the effects of the
relationship between SST and heading angle; and
the third candidate model included the effects of
the relationship between SST and both
components of the empirical tracking model. The
relationships of the translational velocity and
heading angle with SST were considered through
the inclusion of a T
s
term in the equations of the
empirical tracking model as in Eq. 2 and Eq. 3
respectively. .
εθ
χψ
++++
++=
si
TT
Taa
Vaaaa
V
654
321
)ln(
)ln(
(2
)
εθθχψθ
+++++++=
siiT
TbbbVbbbb
71654321
(3
)
where V
T
= translational velocity, ψ = latitude at
eye, χ = longitude at eye, θ = heading, and T
s
=
SST at eye. The Atlantic basin was divided into a
5º square grid, and regression analysis of the
HURDAT data was used to obtain values of the
coefficients a
i
and b
i
for each grid location. For
grid locations with little or no hurricane data, the
coefficients were assigned the values from the
nearest grid location. A unique set of coefficients
was obtained for easterly and westerly moving
5
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
hurricanes. The candidate tracking models
conditioned on SST were used to simulate
10,000 years of hurricanes with initial conditions
(e.g., latitude, longitude, translational velocity,
heading angle, central pressure) in the current
climate.
A comparison of 100 storms simulated in
the current climate using each of the candidate
models versus a subset of historical tracks is
shown in Figure 5. It is clear that the inclusion of
a SST term directly in the computation of the
hurricane heading introduces unrealistic variation
in the track behavior. Landfalling rates
corresponding to Figure 5 are presented in Table
1, for the five segmented regions of the eastern
US coast. Again, it is clear that the candidate
models considering a direct relationship between
SST and the hurricane heading angle are not able
to reproduce historical tracking behavior. From
both Figure 5 and Table 1 however, the
candidate model which only considers the direct
relationship between SST and hurricane
translational velocity (T
S
V
T
) produces results
which agree quite well with the historical data in
overall track behavior as well as in the rate of
landfall on the US coast.
Figure 5: (top-L) Historical tracks of US landfalling
hurricanes (1962-2012); Simulated tracks of 100
storms in the current climate produced from tracking
paramters with SST considerations in relation to
(top-R) translational velocity only, (bottom-L)
heading only, and (bottom-R) translational velocity
and heading.
It should be noted that in this tracking
model, the hurricane heading angle and
translational velocity are coupled. Therefore,
although a direct relationship between SST and
hurricane heading angle was not apparent, effects
of changing SSTs on the hurricane heading angle
are implicitly considered through the effects of
changing SST on the translational velocity.
Table 1: Observed historical rate of landfall and
simulated current climate rate of landfall of
empirical tracking models conditioned on SST for
regions of eastern US coast.
NE
VA-GA
E. FL
GoM
TX
Historical
0.11
0.37
0.35
1.02
0.40
T
S
– V
T
0.12
0.36
0.37
1.08
0.43
T
S
θ
0.08
0.18
0.14
0.56
0.18
T
S
– V
T
, θ
0.10
0.29
0.24
0.76
0.29
4. SIMULATION
The historical rate of landfalling hurricanes for
the northeast US coast was found to be
approximately 0.11 per year. The simulated
landfalling rate in the current climate, for all
model variations, was approximately 0.12 to 0.13
hurricanes per year, for all scenarios.
Considering the effects of SST only on the
hurricane genesis frequency, the annual
landfalling rate in the future climate scenario
approximately doubled to 0.25 hurricanes per
year. The landfalling rate reduced when
considering the effects of SST on hurricane
genesis location and track propagation, to 0.07
and 0.09 hurricanes per year respectively. When
considering the effects of SST on all components
of the hurricane, the simulated landfalling rate in
2100 was approximately 0.08 hurricanes per
year. These results indicate that in a warmed
climate, more events may be produced, but fewer
would actually make landfall in the northeast US.
The focus of this study is to probabilistically
characterize the hurricane hazards at the time of
landfall. To concomitantly characterize the
hurricane hazards, a histogram can be
constructed in four-dimensional space, using
maximum surface wind speed (V
max
), radius of
6
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
maximum winds (R
max
), and rate of rainfall (RR)
recorded at the time of landfall. Bin sizes of 2
m/s, 10 mm/day, and 10 km were used for wind
speed, rainfall intensity, and storm size,
respectively. The annual exceedance probability
is obtained from the histogram by dividing the
number of data points within a specified bin by
the total number of data points, and multiplying
by the annual hurricane occurrence rate. Hazard
levels, with different annual exceedance
probabilities, can then be defined by three-
dimensional equi-probability surfaces. The
hazard level can also be described as an MRI or
as an exceedance probability in N years (e.g.,
2%/50 years). The hazard surfaces for the current
climate scenario and future RCP 8.5 climate
scenario are shown in Figure 6, considering the
relationships between SST and all hurricane
components.
Figure 6: Trivariate hazard level (MRI) surfaces for
the northeast US coast under the current (2012)
climate scenario (left) and the future (2100) RCP 8.5
climate scenario (right).
Concomitantly characterizing the hurricane
hazards, and considering the effects of possible
future changes in SST on the hurricane genesis
frequency, V
max
and RR are projected to increase
13% and 48% respectively, for the 700-year
MRI. This is due to the large number of events
making landfall in the northeast US in this
scenario. Considering the effects of possible
future climate changes in SST only on the
hurricane genesis location (track propagation)
results in less drastic increases in V
max
and RR of
7% and 15% (6% and 22%) respectively, for the
700-year MRI. Including the effects of changes
in SST on all components of the hurricane
simulation, the projected changes in the values of
V
max
and RR for the 700-year MRI under RCP
8.5 are approximately 10% and 50%.
5. CONCLUSIONS
Using state-of-the-art empirical, event-based
hurricane models, an analysis was presented to
investigate possible future climate change impact
on the hurricane wind and rain hazards. Sea
surface temperature served as the index of
climate change herein. Future SSTs were
obtained from projected climate change scenario
RCP 8.5, developed for the IPCC Fifth
Assessment Report. The influence of changes in
SST on the hurricane intensity, size, genesis
frequency, genesis location, and track
propagation were considered separately and
together. A total of 10,000 years of hurricane
events under the current (2012) and future (2050
and 2100) climate scenarios, was simulated to
produce a synthetic hurricane database for every
zip-code in the study region.
The US northeast coastline was used as the
study region. The results of this analysis indicate
that the number of landfalling hurricane events in
the northeast US is likely to decrease in a
warmed climate. However, the hurricane hazards
in the study region were projected to increase at
each design level (e.g., 300, 700, and 1700-year
MRIs) under the RCP 8.5 climate scenario. In
addition, increases in both population and
development along coastal areas are only
expected to increase the vulnerability of this
7
12
th
International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP12
Vancouver, Canada, July 12-15, 2015
region. As seen in recent hurricanes (e.g., Irene,
2011; Sandy, 2012) even moderate hurricanes
are able to have devastating impacts in the
region.
6. REFERENCES
Darling, R. (1991). Estimating probabilities of
hurricane wind speeds using a large scale
empirical model. Journal of Climate, 4 (10),
1035-1046.
Elsner, J. B. (2003). Tracking hurricanes. Bulletin of
the American Meteorological Society, 84 (3),
353-356.
Elsner, J. B., & Jagger, T. H. (2006). Prediction
models for annual US hurricane counts.
Journal of Climate, 19 (12), 2935-2952.
Georgiou, P. N. (1985). Design wind speeds in
tropical cyclone-prone regions (Doctoral
Dissertation). Retrieved from University of
Western Ontario Digitized Theses. (1523).
Hall, T., & Yonekura, E. (2013). North American
tropical cyclone landfall and SST: a statistical
model study. Journal of Climate, 26 (21),
8422-8439.
Hall, T. M., & Jewson, S. (2007). Statistical
modelling of north Atlantic tropical cyclone
tracks. Tellus A, 59 (4), 486-498.
Holland, G. J., & Webster, P. J. (2007). Heightened
tropical cyclone activity in the North Atlantic:
natural variability or climate trend?
Philosophical Transactions of the Royal
Society A: Mathematical, Physical and
Engineering Sciences, 365 (1860), 2695-2716.
Jewson, S., Binter, R., Khare, S., Nzerem, K., &
O'Shay, A. (2008). Predicting basin and
landfalling hurricane numbers from sea
surface temperature. Retrieved June 21, 2013,
from arXiv.org/abs/physics/0701170.
Knutson, T. R., Sirutis, J. J., Garner, S. T., Vecchi, G.
A., & Held, I. M. (2008). Simulated reduction
in Atlantic hurricane frequency under twenty-
first-century warming conditions. Nature
Geoscience, 1 (6), 359-364.
Landsea, C. W. (2000). Climate variability of tropical
cyclones: past, present and future. In Storms
(Vol. 1, pp. 220-241). London, UK: Routledge
Press.
Landsea, C. W., Vecchi, G. A., Bengtsson, L., &
Knutson, T. R. (2010). Impact of duration
thresholds on Atlantic tropical cyclone counts.
Journal of Climate, 23 (10), 2508-2519.
Mann, M. E., & Emanuel, K. A. (2006). Atlantic
hurricane trends linked to climate change.
EOS, Transactions American Geophysical
Union, 87 (24), 233-241.
Mann, M. E., Sabbatelli, T. A., & Neu, U. (2007).
Evidence for a modest undercount bias in early
historical Atlantic tropical cyclone counts.
Geophysical Research Letters, 34 (22), 1-6.
Mudd, L., Wang, Y., Letchford, C., & Rosowsky, D.
(2014). Assessing Climate Change Impact on
the US East Coast Hurricane Hazard:
Temperature, Frequency, and Track. Natural
Hazards Review, 15 (3), 1-13.
Pachauri, R. K. (2014). Climate change 2014.
synthesis report. contribution of working
groups I, II and III to the fifth assessment
report. Geneva, CH: Intergovernmental Panel
on Climate Change.
Vecchi, G. A., & Knutson, T. R. (2008). On estimates
of historical north Atlantic tropical cyclone
activity. Journal of Climate, 21 (14), 3580-
3600.
Vickery, P. J., Skerlj, P., Steckley A. & Twisdale, L.
(2000a). Hurricane wind field model for use in
hurricane simulations. Journal of Structural
Engineering, 126 (10), 1203-1221.
Vickery, P. J., Skerlj, P., & Twisdale, L. (2000b).
Simulation of hurricane risk in the US using
empirical track model. Journal of Structural
Engineering, 126 (10), 1222-1237.
8
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
This paper presents a study to assess the impact of possible future climate change on the hurricane wind hazard along the US eastern coastline. Initially, climate change scenarios were coupled with state-of-the-art hurricane genesis, wind field, and tracking models to examine possible changes in hurricane intensity (maximum wind speed) and hurricane size (radius to maximum winds). A number of different postulated climate change models (IPCC scenarios) were considered. Each scenario suggested changes in sea surface temperature (SST), which is the driving parameter in most modern hurricane models. The evolution of hurricane genesis frequency was then considered both independently and jointly with hurricane intensification. State-of-the-art probabilistic event modeling and simulation techniques were used to generate 10,000 years of hurricane events under the 2005 and future climate conditions. The annual maximum wind speed distribution and the joint distribution of maximum wind speed and storm size, under 2005 and future climate scenarios, are then compared. Finally, the evolution of hurricane tracks was examined, in an effort to establish a trend over time.
Article
Full-text available
This paper presents a study to assess the impact of possible future climate change on the hurricane wind hazard along the eastern coastline of the United States. Initially, climate change scenarios were coupled with state-of-the-art hurricane genesis, wind field, and tracking models to examine possible changes in hurricane intensity (maximum wind speed) and hurricane size (radius to maximum winds). A number of different postulated climate change models (IPCC scenarios) were considered. Each scenario suggested changes in sea surface temperature (SST), which is the driving parameter in most modern hurricane models. The evolution of hurricane genesis frequency was then considered both independently and jointly with hurricane intensification. State-of-the-art probabilistic event modeling and simulation techniques were used to generate 10,000 years of hurricane events under the 2005 and future climate conditions. The annual maximum wind speed distribution and the joint distribution of maximum wind speed and storm size, under 2005 and future climate scenarios, are then compared. Finally, the evolution of hurricane tracks was examined in an effort to establish a trend over time. (C) 2014 American Society of Civil Engineers.
Article
Full-text available
Increases in key measures of Atlantic hurricane activity over recent decades are believed to reflect, in large part, contemporaneous increases in tropical Atlantic warmth [e.g., Emanuel, 2005]. Some recent studies [e.g., Goldenberg et al., 2001] have attributed these increases to a natural climate cycle termed the Atlantic Multidecadal Oscillation (AMO), while other studies suggest that climate change may instead be playing the dominant role [Emanuel, 2005; Webster et al., 2005]. Using a formal statistical analysis to separate the estimated influences of anthropogenic climate change from possible natural cyclical influences, this article presents results indicating that anthropogenic factors are likely responsible for long-term trends in tropical Atlantic warmth and tropical cyclone activity. In addition, this analysis indicates that late twentieth century tropospheric aerosol cooling has offset a substantial fraction of anthropogenic warming in the region and has thus likely suppressed even greater potential increases in tropical cyclone activity.
Article
A statistical-stochastic model of the complete life cycle of North Atlantic (NA) tropical cyclones (TCs) is used to examine the relationship between climate and landfall rates along the North American Atlantic and Gulf Coasts. The model draws on archived data of TCs throughout the North Atlantic to estimate landfall rates at high geographic resolution as a function of the ENSO state and one of two different measures of sea surface temperature (SST): 1) SST averaged over theNAsubtropics and the hurricane season and 2) this SST relative to the seasonal global subtropicalmean SST (termed relSST).Here, the authors focus on SST by holding ENSOto a neutral state. Jackknife uncertainty tests are employed to test the significance of SST and relSST landfall relationships. There are more TC and major hurricane landfalls overall in warm years than cold, using either SST or relSST, primarily due to a basinwide increase in the number of storms. The signal along the coast, however, is complex. Some regions have large and significant sensitivity (e.g., an approximate doubling of annualmajor hurricane landfall probability on Texas from -2 to+2 standard deviations in relSST), while other regions have no significant sensitivity (e.g., the U.S. mid-Atlantic and Northeast coasts). This geographic structure is due to both shifts in the regions of primary TC genesis and shifts in TC propagation.
Article
A critical component,in the simulation of hurricanes,is a good,representation of the hurricane wind field when given information of the storm intensity, size, and translation speed. In the investigation de- scribed here, the full nonlinear solution to the equations of motion of a hurricane are solved and then parame- terized for use in fast-running simulations. The hurricane model,described here takes into account the effects of changing,sea surface roughness,and the air-sea temperature,difference on the estimated surface-level wind speeds. Comparisons,between,modeled,and observed,hurricane wind,speed records are performed,where,one compares both the 10-min mean,wind,speeds,and the peak,gust wind,speeds. The resulting wind,field model,represents the most,physically based and validated model,used in the estimate of hurricane wind,speed exceedance,prob- abilities.
Article
There is widespread concern about the recent increase in North Atlantic hurricane activity. Results here suggest that fledgling storms tracking east to west at low latitudes are more likely to reach hurricane intensity than those traveling on a more northerly trajectory. The annual occurrence of these straight-moving hurricanes (east to west at low latitudes) is statistically linked to the El Niño-Southern Oscillation (ENSO) and to the North Atlantic Oscillation (NAO) using a Poisson regression. Because the occurrence of U.S. hurricanes south of about 35°N is positively correlated with the abundance of straight-moving hurricanes, an accurate prediction of ENSO together with observations of the NAO could be used to forecast seasonal hurricane probabilities along the southeast U.S. coast. It is stressed that in order to understand the range of mechanisms associated with hurricane activity, it is important to consider factors that influence tracks. In this regard, the NAO is a leading candidate.
Article
A new method is presented for estimating the probability of exceeding a given wind speed in 1 year at any given location in the Atlantic tropical cyclone basin. The method is especially appropriate for wind speeds with return periods of 100 years or more, for which on-the-spot data are inadequate. The relative intensity of a tropical cyclone at any point in time is the actual central pressure drop divided by the greatest possible central pressure drop that mean seasonal climatic conditions allow. The empirical distribution of relative intensity as a function of time is derived from all Atlantic cyclone data. By combining this with information about the distribution of time from cyclone inception to closest approach to the site, and other steps, probabilities for various wind speeds at the site are obtained. Unlike other methods, this one does not attempt to fit data to any extreme value distribution, and it performs all integrations by discretizing and summing explicitly, never by simulation. The results at the Turkey Point Power Plant site in south Florida am in good agreement at lower wind speeds with National Hurricane Center estimates using local data for 1886-1988.