Trends in global tropical cyclone activity over the past twenty years
(1986–2005)
Philip J. Klotzbach
1
Received 3 February 2006; revised 2 March 2006; accepted 18 April 2006; published 20 May 2006.
[
1
]
The recent destructive Atlantic hurricane seasons and
several recent publications have sparked debate over
whether warming tropical sea surface temperatures (SSTs)
are causing more intense, longer-lived tropical cyclones.
This paper investigates worldwide tropical cyclone
frequency and intensity to determine trends in activity
over the past twenty years during which there has been an
approximate 0.2
°
– 0.4
°
C warming of SSTs. The data
indicate a large increasing trend in tropical cyclone
intensity and longevity for the North Atlantic basin and a
considerable decreasing trend for the Northeast Pacific. All
other basins showed small trends, and there has been no
significant change in global net tropical cyclone activity.
There has been a small increase in global Category 4 – 5
hurricanes from the period 1986 – 1995 to the period 1996 –
2005. Most of this increase is likely due to improved
observational technology. These findings indicate that other
important factors govern intensity and frequency of tropical
cyclones besides SSTs.
Citation:
Klotzbach, P. J. (2006),
Trends in global tropical cyclone activity over the past twenty
years (1986 – 2005),
Geophys. Res. Lett.
,
33
, L10805, doi:10.1029/
2006GL025881.
1.
Introduction
[
2
] Recent papers by
Emanuel
[2005] and
Webster et al.
[2005] have caused a flurry of debate about the relationship
between increasing tropical sea surface temperatures (SSTs)
and intense tropical cyclones (TCs).
Emanuel
[2005] found
that a Power Dissipation Index (PDI), effectively the six-
hour TC one-minute maximum sustained wind speed cubed,
had increased by approximately 50% for both the Atlantic
basin and the Northwest Pacific basin since the mid 1970s.
Webster et al.
[2005] analyzed Category 4 – 5 hurricanes
(maximum sustained winds (1-minute average) >= 115
knots) for all TC basins over the past 30 years and found
that their numbers had nearly doubled between an earlier
(1975 – 1989) and a more recent (1990 – 2004) 15-year
period. Many questions have been raised regarding the data
quality in the earlier part of their analysis periods (C. W.
Landsea et al., Global warming and extreme tropical cyclo-
nes: Can we detect climate trends from existing tropical
cyclone databases?, submitted to
Science
, 2006). Before the
early 1980s, the Dvorak Technique [
Dvorak
, 1975], a
method which utilizes satellite imagery to assign an inten-
sity to TCs, was only applicable to visible satellite imagery
and therefore could not be used at night. Since 1984,
improved technology has allowed the technique to be
applied to both infrared and visible imagery [
Dvorak
,
1984], and more accurate estimates of real-time intensity
have become available. In addition, the quality and resolu-
tion of satellite imagery has continued to improve over time,
and with this improved imagery, operational forecasters can
be more confident of their satellite-derived intensity esti-
mates. The elimination of aircraft reconnaissance in the
Northwest Pacific in 1987 raised the importance of satellite-
based intensity estimates even more. Also, the Joint Ty-
phoon Warning Center urges caution in utilizing data prior
to 1985 [
Chu et al.
, 2002]. Because of these earlier period
limitations and the desire to obtain a near-homogeneous
data set, only the past twenty years (1986 – 2005) are
examined in this paper. If the trends shown by
Emanuel
[2005] and
Webster et al.
[2005] are to be accepted, then
one should also find a similar increasing trend in global TC
data sets over the last 20 years.
2.
Methodology
[
3
] Global TC activity was tabulated using ‘‘best track’’
data sets from 1986 – 2004 for all TC basins (the North
Atlantic, the Northeast Pacific, the Northwest Pacific, the
North Indian, the South Indian, and the South Pacific). The
‘‘best track’’ data sets are the best estimates of the locations
and intensities of TCs at six-hour intervals produced by the
international warning centers. The ‘‘best track’’ data sets
from the National Hurricane Center (NHC) were utilized for
the North Atlantic [
Jarvinen et al.
, 1984] and the Northeast
Pacific basins, and ‘‘best track’’ data from the Joint Ty-
phoon Warning Center (JTWC) [
Chu et al.
, 2002] were
utilized for the North Indian and Northwest Pacific basins.
[
4
] For the South Indian and South Pacific basins, a data
set created by Neumann [
Neumann
, 1999] was used for
1986 – 2001 because it was utilized by
Webster et al.
[2005]
in their study. The South Indian and South Pacific basins
were divided at 135
°
E with storms forming east of this
longitude being classified as South Pacific storms and storms
forming west of this longitude being classified as South
Indian storms. If a storm crossed 135
°
E longitude, it was
classified into the basin in which it accrued more named
storm days. The Neumann data set ended in June 2002, and
after this point, the JTWC’s ‘‘best track’’ data set was used.
The JTWC data set overlaps the Neumann data set from
1970 – 2002, and the correlation between TC statistics cal-
culated from these data sets is greater than 0.95. The
consistency between data sets suggests that the JTWC ‘‘best
track’’ data set can be utilized from July 2002 – June 2004
without causing any spurious jumps in the data.
[
5
] For 2005 for the Northern Hemisphere and for July
2004 – 2005 for the Southern Hemisphere, operational TC
GEOPHYSICAL RESEARCH LETTERS, VOL. 33, L10805, doi:10.1029/2006GL025881, 2006
1
Department of Atmospheric Science, Colorado State University, Fort
Collins, Colorado, USA.
Copyright 2006 by the American Geophysical Union.
0094-8276/06/2006GL025881$05.00
L10805
1 of 4
intensity estimates were utilized. These data were obtained
from the NHC for the North Atlantic and the Northeast
Pacific. JTWC data were utilized for all other basins. In the
Southern Hemisphere, JTWC advisories were occasionally
supplemented with data from the advisory centers in Perth,
Darwin, Brisbane and La Reunion to extend a storm’s
length or increase its intensity slightly per the suggestion
of Gary Padgett (personal communication, 2006). This was
done for storms where the JTWC advisory intensities were
considerably below the intensity recorded at the other
centers. The combination of these data sets provides a
comprehensive evaluation of global TC activity over the
past twenty years (1986 – 2005).
3.
Trends in Accumulated Cyclone Energy
[
6
] Figure 1 displays Accumulated Cyclone Energy
(ACE) index values for all TC basins from 1986 – 2005.
Linear trends have been fitted to all six TC basins. ACE is
defined to be the sum of the maximum one-minute sus-
tained surface wind speed squared at six-hourly intervals for
all periods when the TC is at least of tropical storm strength
(>= 34 knots) [
Bell et al.
, 2000]. ACE is proportional to the
kinetic energy generated by the storm. This ACE index is
quite similar to the Power Dissipation Index (PDI) created
by
Emanuel
[2005], since PDI is defined to be the sum of
the maximum one-minute sustained wind speed cubed at
six-hourly intervals for all periods when the TC is at least of
tropical storm strength. PDI is roughly proportional to the
amount of monetary damage or power dissipation generated
by a tropical cyclone [
Emanuel
, 2005]. In this paper, ACE is
used instead of PDI as changes proportional to kinetic
energy are being evaluated. Trends would be virtually the
same using the PDI index, as the ACE and PDI indices
correlate globally at 0.97. The largest trends in ACE
noticeable in this figure are a large increase over the past
twenty years in the North Atlantic and a considerable
decrease over the Northeast Pacific. The large recent in-
crease in North Atlantic activity has been noted extensively
throughout the literature and has been attributed to an
increase in strength of the Atlantic Thermohaline Circula-
tion (THC) (alternatively referred to as a change in sign to a
positive phase of the Atlantic multidecadal mode) [
Gray et
al.
, 1997;
Goldenberg et al.
, 2001;
Klotzbach and Gray
,
2005;
Pielke et al.
, 2005]. The trends in all other basins are
quite small.
[
7
] Figure 2 takes the ACE index values for all TC basins
and sums them into values for the Northern Hemisphere, the
Southern Hemisphere, and the entire globe. A five-year
running mean of tropical SST anomalies (23.5
°
N – 23.5
°
S,
all longitudes) obtained from the NCEP Reanalysis [
Kistler et
al.
, 2001] are also plotted for reference. A linear trend has
been fitted to all three curves, and it is noted that there is a
slight increase in ACE for the Northern Hemisphere, the
Southern Hemisphere and consequently for the globe for the
1986 – 2005 period. As seen in Figure 3, if the last 16 years of
the data set are examined (1990 – 2005), the trend in global
ACE is actually slightly downward, although tropical SSTs
increased by approximately 0.2
°
– 0.3
°
C during this period.
[
8
] Table 1 displays total ACE for the ten-year periods of
1986 – 1995 and 1996 – 2005 for each of the individual TC
basins as well as for the combination of the North Atlantic
and Northeast Pacific, the Northern Hemisphere, the South-
ern Hemisphere and for all TCs worldwide. Ratios of the
second ten-year period to the first ten-year period are
calculated. Average tropical SSTs for each ten-year period
are provided for reference. Effectively, when grouped into
ten-year periods, there has been virtually no trend in globally-
Figure 1.
Accumulated Cyclone Energy (ACE) index
values for individual TC basins from 1986 – 2005.
Figure 2.
Accumulated Cyclone Energy (ACE) index
values for 1986 – 2005 for the Northern Hemisphere (NH)
(green line), the Southern Hemisphere (SH) (pink line) and
the globe (brown line). The dashed lines are linear trends
that have been fitted to the three curves. Five-year running
mean tropical NCEP Reanalysis SST anomalies (23.5
°
N –
23.5
°
S, all longitudes) (blue line) are also plotted. The base
period for tropical SSTs is 1951 – 1980.
L10805
KLOTZBACH: TRENDS IN GLOBAL TC ACTIVITY (1986 – 2005)
L10805
2 of 4
summed ACE. The largest individual-basin trends are evident
in the North Atlantic and the Northeast Pacific. There has
been a large increase in ACE over the past decade in the North
Atlantic, and there has been a large decrease in ACE over the
past decade in the Northeast Pacific. When ACE in the North
Atlantic and Northeast Pacific are added together, there has
been a very small increase in Western Hemisphere TC activity
over the past twenty years. A slight negative trend in ACE is
noted in the Northwest Pacific, which is in contrast to the large
increase in PDI noted by
Emanuel
[2005] since the mid
1970s. One would expect the increasing trend in PDI over
the past thirty years to also show an increase over the past
twenty years when sea surface temperatures have warmed by
approximately 0.2
°
– 0.4
°
C.
4.
Trends in Category 4– 5 Hurricanes
[
9
]
Webster et al.
[2005] report that there has been a large
(nearly 100%) increase in Category 4 – 5 hurricanes since the
mid 1970s. Table 2 displays the number of Category 4 – 5
hurricanes by individual TC basin, for the North Atlantic and
the Northeast Pacific, for the Northern Hemisphere, the
Southern Hemisphere and the globe by ten-year periods since
1986. Northern Hemisphere Category 4 – 5 hurricanes have
remained virtually the same between the two ten-year periods,
and a modest increase in Category 4 – 5 hurricanes has been
observed in the Southern Hemisphere. Most of this Southern
Hemisphere increase occurred in the first five years of the data
set, and since the early 1990s, as satellite observational
technology has continued to improve, there has been no
continuation of this trend; even though global SSTs and
oceanic heat content have continued to rise [
Levitus et al.
,
2000]. During the past twenty years, the number of Category
4 – 5 hurricanes has increased dramatically in the North
Atlantic, and there has been a large decrease in the Northeast
Pacific, in keeping with the ACE values for these basins.
Since 1990, the number of Category 4 – 5 hurricanes across
the globe has remained approximately constant which agrees
with the findings of
Webster et al.
[2005, Figure 4] who show
effectively no trend in Category 4 – 5 hurricanes from the
1990 – 1994 pentad to the 2000 – 2004 pentad.
5.
Correlations between ACE and Category 4– 5
Hurricanes with Sea Surface Temperatures
[
10
] Table 3 shows the correlation between ACE and SSTs
along with the correlation between Category 4 – 5 hurricanes
and SSTs for each TC basin over the period 1986 – 2005 for
the Northern Hemisphere and for 1985 – 1986 through the
2004 – 2005 hurricane seasons for the Southern Hemisphere.
SSTs are taken from the Hadley SST data set [
Rayner et al.
,
2003]. The Hadley SST data set was correlated from 1986 –
2005 with the NCEP Reanalysis SST data set [
Kistler et al.
,
2001] and the Kaplan SST data set [
Kaplan et al.
, 1998], and
both the NCEP Reanalysis and Kaplan SST data sets corre-
lated at greater than 0.90 with Hadley SSTs for each TC basin.
Therefore, the correlation would not change much if another
SST data set were selected. TC basins are defined as in
Webster et al.
[2005].
[
11
] Based on theoretical research [
Emanuel
, 1987], one
would expect there to be a positive correlation between SST in
Figure 3.
Global Accumulated Cyclone Energy (ACE)
index values for 1990 – 2005 (brown line). A linear trend
has been fitted to global ACE. Five-year running mean
tropical NCEP Reanalysis SST anomalies (23.5
°
N – 23.5
°
S,
all longitudes) (blue line) are also plotted. The base period
for tropical SSTs is 1951 – 1980.
Table 1.
Accumulated Cyclone Energy (Ace) Index Values for
Ten-Year Periods (1986 – 1995, 1996 – 2005) and the Ratio of the
Second Ten-Year Period to the First Ten-Year Period for all TC
Basins, the North Atlantic and the Northeast Pacific, the Northern
Hemisphere, the Southern Hemisphere, and the Globe
a
Basin
1986 – 1995
1996 – 2005
Ratio
(1996 – 2005/
1986 – 1995) %
North Atlantic
762
1438
189%
Northeast Pacific
1646
1037
63%
N. Atlantic + NE Pacific
2408
2475
103%
Northwest Pacific
3495
3307
95%
North Indian
123
180
146%
South Indian
1377
1456
106%
South Pacific
757
755
100%
Northern Hemisphere
6026
5962
99%
Southern Hemisphere
2134
2211
104%
Global
8160
8173
100%
Tropical SSTs,
23.5
°
N – 23.5
°
S,
all longitudes
0.18
°
C
0.29
°
C
D
T = +0.11
°
C
a
Tropical SSTs for each ten-year period (23.5
°
N – 23.5
°
S, all longitudes)
derived from the NCEP Reanalysis and the difference between these two
periods are provided; the base period for tropical SSTs is 1951 – 1980.
Table 2.
Category 4 – 5 Hurricanes by Ten-Year Periods (1986 –
1995, 1996 – 2005) for Individual TC Basins, the North Atlantic
and the Northeast Pacific, the Northern Hemisphere, the Southern
Hemisphere, and the Globe
Basin
1986 – 1995
1996 – 2005
Ratio
(1996 – 2005/
1986 – 1995)
North Atlantic
10
25
250%
Northeast Pacific
37
23
62%
N. Atlantic + NE
Pacific
47
48
102%
Northwest Pacific
75
76
101%
North Indian
3
4
133%
South Indian
26
36
138%
South Pacific
13
16
123%
Northern Hemisphere
125
128
102%
Southern Hemisphere
39
52
133%
Global
164
180
110%
L10805
KLOTZBACH: TRENDS IN GLOBAL TC ACTIVITY (1986 – 2005)
L10805
3 of 4
a TC basin and observed TC intensity, and therefore a one-
tailed Student’s t-test was used to test for statistical signifi-
cance. Since there are 20 years of data, a correlation of 0.38 is
needed to be statistically significant at the 95% level. There is
a statistically significant relationship between SSTs and ACE
as well as SSTs and Category 4 – 5 hurricanes for both the
North Atlantic and the Northeast Pacific; however, correla-
tions for the other four TC basins are actually slightly
negative. Even for the North Atlantic and the Northeast
Pacific, these correlations only explain between 25 – 30% of
the variance, and therefore large amounts of variance are
unexplained. Clearly, other atmospheric and oceanic features
such as vertical wind shear, mid-level instability, etc. [e.g.,
Gray
, 1968] are critical for TC development and intensifica-
tion besides warm SSTs.
6.
Conclusions
[
12
] These findings are contradictory to the conclusions
drawn by
Emanuel
[2005] and
Webster et al.
[2005].
They do not support the argument that global TC
frequency, intensity and longevity have undergone
increases in recent years. Utilizing global ‘‘best track’’
data, there has been no significant increasing trend in
ACE and only a small increase (
10%) in Category 4 – 5
hurricanes over the past twenty years, despite an increase
in the trend of warming sea surface temperatures during
this time period.
[
13
] The results of this paper are more in line with a prior
study by
Shapiro and Goldenberg
[1998] and a project report
by
Gray and Klotzbach
[2005].
Shapiro and Goldenberg
[1998] showed only marginally significant correlations
between SSTs in the tropical Atlantic and major hurricane
development in the basin. Vertical wind shear was shown to
be a much more fundamental component for major hurri-
cane development and maintenance.
Gray and Klotzbach
[2005], while developing seasonal hurricane forecasts
for TC activity, found only a modest correlation (
0.4)
between seasonal and monthly Atlantic basin SSTs and
major (Category 3 – 4 – 5) hurricane frequency. This study
indicates that, based on data over the last twenty years, no
significant increasing trend is evident in global ACE or in
Category 4 – 5 hurricanes.
[
14
]
Acknowledgments.
I would like to acknowledge funding pro-
vided by NSF Grant ATM-0346895 and by the Research Foundation of
Lexington Insurance Company (a member of the American International
Group). Valuable feedback on an earlier version of this manuscript was
provided by William Gray, Jonathan Vigh, and Brian McNoldy. I would
also like to thank Gary Padgett for help in obtaining some of the 2005
operational data. The comments by the reviewers are gratefully acknowl-
edged for helping to improve the quality of the manuscript.
References
Bell, G. D., M. S. Halpert, R. C. Schnell, R. W. Higgins, J. Lawrimore,
V. E. Kousky, R. Tinker, W. Thiaw, M. Chelliah, and A. Artusa (2000),
Climate assessment for 1999,
Bull. Am. Meteorol. Soc.
,
81
(6), 1328.
Chu, J.-H., C. R. Sampson, A. S. Levin, and E. Fukada (2002), The Joint
Typhoon Warning Center tropical cyclone best tracks 1945 – 2000, report,
Joint Typhoon Warning Cent., Pearl Harbor, Hawaii.
Dvorak, V. F. (1975), Tropical cyclone intensity and forecasting from sa-
tellite images,
Mon. Weather Rev.
,
103
, 420 – 430.
Dvorak, V. F. (1984), Tropical cyclone intensity analysis using satellite
data,
NOAA Tech. Rep. NESDIS 11
, 47 pp., NOAA/NESDIS, Washing-
ton, D. C.
Emanuel, K. A. (1987), The dependence of hurricane intensity on climate,
Nature
,
326
, 483 – 485.
Emanuel, K. A. (2005), Increasing destructiveness of tropical cyclones over
the past 30 years,
Nature
,
326
, 686 – 688.
Goldenberg, S. B., C. W. Landsea, A. M. Mestas-NunËœez, and W. M. Gray
(2001), The recent increase in Atlantic hurricane activity: Causes and
implications,
Science
,
293
, 474 – 479.
Gray, W. M. (1968), Global view of the origin of tropical disturbances and
storms,
Mon. Weather Rev.
,
96
, 669 – 700.
Gray, W. M., and P. J. Klotzbach (2005), Summary of 2005 Atlantic tropical
cyclone activity and verification of author’s seasonal and monthly fore-
casts, report, 48 pp., Dep. of Atmos. Sci., Colorado State Univ., Fort
Collins, Colo.
Gray, W. M., J. D. Sheaffer, and C. W. Landsea (1997), Climate trends
associated with multi-decadal variability of intense Atlantic hurricane
activity, in
Hurricanes
,
Climatic Change and Socioeconomic Impacts:
A Current Perspective
, edited by H. F. Diaz and R. S. Pulwarty,
pp. 15 – 53, Springer, New York.
Jarvinen, B. R., C. J. Neumann, and M. A. S. Davis (1984), A tropical
cyclone data tape for the North Atlantic basin, 1886 – 1983: Contents,
limitations, and uses,
Tech. Memo. NWS NHC-22
, 21 pp., NOAA, Wa-
shington, D. C.
Kaplan, A., M. A. Cane, Y. Kushnir, A. C. Clement, M. B. Blumenthal, and
B. Rajagopalan (1998), Analyses of global sea surface temperature
1856 – 1991,
J. Geophys. Res.
,
103
(C9), 18,567 – 18,589.
Kistler, R., et al. (2001), The NCEP – NCAR 50 – year reanalysis: Monthly
means CD – ROM and documentation,
Bull. Am. Meteorol. Soc.
,
82
(2),
247 – 267.
Klotzbach, P. J., and W. M. Gray (2005), Extended range forecast of Atlan-
tic seasonal hurricane activity and U.S. landfall strike probability for
2006, report, 24 pp., Dep. of Atmos. Sci., Colorado State Univ., Fort
Collins, Colo.
Levitus, S., J. I. Antonov, T. P. Boyer, and C. Stephens (2000), Warming of
the world ocean,
Science
,
287
, 2225 – 2229.
Neumann, C. J. (1999), The HURISK model: An adaptation for the South-
ern Hemisphere (A user’s manual), report, contract N00014-96-C-6015,
31 pp., Sci. Appl. Int. Corp., Monterey, Calif.
Pielke, R. A., Jr., C. Landsea, M. Mayfield, J. Laver, and R. Pasch (2005),
Hurricanes and global warming,
Bull. Am. Meteorol. Soc.
,
86
(11), 1571 –
1575.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander,
D. P. Rowell, E. C. Kent, and A. Kaplan (2003), Global analyses of sea
surface temperature, sea ice, and night marine air temperature since the
late nineteenth century,
J. Geophys. Res.
,
108
(D14), 4407, doi:10.1029/
2002JD002670.
Shapiro, L. J., and S. B. Goldenberg (1998), Atlantic sea surface tempera-
tures and tropical cyclone formation,
J. Clim.
,
11
, 578 – 590.
Webster, P. J., G. J. Holland, J. A. Curry, and H.-R. Chang (2005), Changes
in tropical cyclone number and intensity in a warming environment,
Science
,
309
, 1844 – 1846.
P. J. Klotzbach, Department of Atmospheric Science, Colorado State
University, Fort Collins, CO 80523, USA. (philk@atmos.colostate.edu)
Table 3.
Correlations Between ACE and SSTs and Category 4 – 5
Hurricanes and SSTs for All TC Basins
a
Basin
Correlation
with ACE
Correlation with
Cat. 4 – 5 Hurricanes
North Atlantic
0.57
0.39
Northeast Pacific
0.58
0.59
Northwest Pacific
0.28
0.11
North Indian
0.07
0.29
South Indian
0.32
0.18
South Pacific
0.38
0.20
a
TC basins and seasons are defined by
Webster et al.
[2005] (North
Atlantic Ocean - 5
°
to 25
°
N, 20
°
to 90
°
W, June – October), (Northeast
Pacific Ocean - 5
°
to 20
°
N, 90
°
to 120
°
W, June – October), (Northwest
Pacific Ocean - 5
°
to 20
°
N, 120
°
to 180
°
E, May – December), (North Indian
Ocean - 5
°
to 20
°
N, 55
°
to 90
°
E, April – May and September – November),
(South Indian Ocean - 5
°
to 20
°
S, 50
°
to 115
°
E, November – April), and the
(South Pacific Ocean - 5
°
to 20
°
S, 155
°
to 180
°
E, December – April);
correlations significant at the 95% level based on a one-tailed Student’s t-test
are bold-faced.
L10805
KLOTZBACH: TRENDS IN GLOBAL TC ACTIVITY (1986 – 2005)
L10805
4 of 4