By: Christopher W. Landsea
Department of Atmospheric Science
Colorado State University
Fort Collins, Colorado 80523

Monthly Weather Review, Vol.121, pg. 1703-1713, (1993)
(Manuscript received 23 March 1992, in final form 12 October 1992)

AMS Copyright Notice

© Copyright 199x American Meteorological Society (AMS). Permission to use figures, tables, and brief excerpts from this work in scientific and educational works is hereby granted provided that the source is acknowledged. Any use of material in this work that is determined to be "fair use" under Section 107 or that satisfies the conditions specified in Section 108 of the U.S. Copyright Law (17 USC, as revised by P.L. 94-553) does not require the Society's permission. Republication, systematic reproduction, posting in electronic form on servers, or other uses of this material, except as exempted by the above statements, requires written permission or license from the AMS. Additional details are provided in the AMS Copyright Policies, available from the AMS at 617-227-2425 or amspubs@ametsoc.org. Permission to place a copy of this work on this server has been provided by the AMS. The AMS does not guarantee that the copy provided here is an accurate copy of the published work.


The variability of intense (or major) hurricanes in the Atlantic basin is investigated on both intraseasonal and interannual timescales. Differences are highlighted in characteristics between intense hurricanes and the weaker minor hurricanes/tropical storms. Intense hurricanes show a much more peaked annual cycle than do weaker tropical cyclones. Ninety-five percent of all intense hurricane activity occurs during August to October. In addition, over eighty percent of all intense hurricanes originate from African easterly waves, a much higher proportion than is observed for weaker cyclones. Of all classes of Atlantic basin tropical cyclones, the intense hurricanes display the greatest year-to-year variability. The incidence of intense hurricanes also has decreased during the last two decades. A small portion of this decreased activity appears to be due to an overestimation of hurricane intensity during the period spanning the 1940s through the 1960s. However, after adjusting for this bias, a substantial downward trend in intense hurricane activity during recent years is still apparent. Given that intense hurricanes are responsible for more than seventy percent of all destruction caused by tropical cyclones in the United States, an understanding is needed of the physical mechanisms for these observed variations of intense hurricane activity.


Climate change due to an increase of anthropogenic ``greenhouse gases'' may warm tropical sea surface temperatures although the magnitude, timing, and spatial variations of such a warming are very uncertain (IPCC 1990). One hypothesis is that increased sea surface temperatures will cause ``a higher frequency and greater intensity of hurricanes" (AMS Council and UCAR Board of Trustees 1988). Gray (1989), however, has suggested that warmer sea surface temperatures do not necessarily mean increased hurricane activity. He stated that ``scenarios can just as easily be made for reductions in hurricane activity due to greenhouse influences.''

In either case, before extrapolations to future frequencies of intense hurricanes can be attempted, it is essential that there be an understanding of the current annual cycle and interannual variability of intense hurricane occurrence. Intense (or major) hurricanes (IH) are defined as those tropical cyclones with maximum sustained (1 min) surface winds of 50 m s-1 during some part of their lifetimes (Hebert and Taylor 1978). These are the Saffir-Simpson (Simpson 1974) category 3, 4, and 5 hurricanes (Table 1).

Previous climatologies of Atlantic basin (i.e., North Atlantic Ocean, Caribbean Sea, and Gulf of Mexico) tropical cyclones by Mitchell (1924), Cry (1965), and Neumann et al. (1987) have primarily focused on the seasonal numbers of named storms (i.e., storms with maximum sustained wind speeds of 18 m s-1) and hurricanes (33 m s-1). Hebert and Taylor (1978), Hebert and Case (1990), and Hebert et al. (1992) discussed IH which have affected the United States, including details of U.S. IH frequency and U.S. IH spawned death and destruction. However, no work has been done on basinwide IH activity for either the intrannual or interannual timescales.

This paper partly fills this void by providing discussions of the annual cycle of IH activity, the interannual climatology of Atlantic basin IH since the mid-1940s and the interannual variations in the origins of IH, as well as providing updates on the U.S. IH frequency and on the IH contribution to U.S. tropical cyclone spawned damage.


a. Data Sources

Most of the data sets utilized for this study are detailed in Landsea and Gray (1992). The Atlantic basin ``Best Track" (Jarvinen et al. 1984) data set includes the six-hour positions, estimated maximum sustained surface winds, and any measured central pressures of tropical cyclones from 1886 to 1991. The second data set which includes U.S. landfalling hurricanes since 1899 categorized by intensity with the Saffir-Simpson Scale was first documented by Hebert and Taylor (1975). Circulation features associated with the origins of all Atlantic basin tropical cyclones since 1967 were provided by Avila (personal communication). In addition, variations in dollar values for tropical cyclone spawned damage in the United States were studied utilizing the yearly summaries appearing in the Monthly Weather Review from 1944 to 1991. Changes in some of the damage figures are given in Hebert and Case (1990).

b. Data Descriptions and Limitations

As stated previously, no climatology has adequately described basinwide IH variability. Part of this neglect may be due to a lack of confidence in analyzed tropical cyclone intensity (as measured by maximum sustained surface winds or by the lowest central pressure) until recent years. The U.S. Air Force and Navy began the practice of routine reconnaissance flights into the storms in 1944 (see Sumner 1944). Before then, unless a tropical cyclone went directly over a ship or a coastal station, the exact intensity of the cyclone was unknown. Of course, not every tropical cyclone in the Atlantic basin since 1944 has received aircraft monitoring every six hours of its existence. However, as described by Clark (1960), tropical cyclones which may threaten coastal areas are kept under nearly continuous aircraft surveillance. This is still true, as was the case for 1991's Hurricane Bob (McAdie and Rappaport 1991).

There is also the question of incidence of unobserved storms during the early days of hurricane forecasting. Certainly, before 1944 a number of short-lived tropical cyclones in the eastern and central Atlantic were likely completely missed. The monitoring situation improved greatly with the advent of aircraft reconnaissance. Nonetheless, while synoptic flights were sent out regularly into the central Atlantic before reports of tropical cyclones were received, some tropical cyclones may still have gone undetected in the pre-satellite, post-aircraft era. However, these are likely few in number and weak in intensity.

Because of these uncertainties, Neumann et al. (1987) suggest utilizing tropical cyclone statistics since 1944 as this period probably best represents the frequencies and intensities of tropical cyclones in the Atlantic. For this reason, the Atlantic basin interannual variability study includes only Best Track data after 1943, a severe limitation for analysis of daily incidence of activity because of the relative scarcity of IH in the Atlantic basin relative to the more active Northwest Pacific basin. The intraseasonal statistics, however, are based upon the entire Best Track data set from 1886 to 1991. Although the data from 1886 to 1943 are incomplete because of unobserved and misclassified tropical cyclones, there is likely no systematic intraseasonal bias induced in using the longer term record.

Reliable data are available, however, for hurricanes which have affected the United States since 1899. There are a few minor inconsistencies, though, between this data set and the Best Track because of slight differences in classification methodology [see Hebert and Taylor (1975) and Jarvinen et al. (1984)]. However, these small differences do not affect the conclusions in this paper.

Before continuous satellite coverage in the mid-1960s, identification of the storm formation characteristics of individual tropical cyclones is unreliableexcept for a few storms. Starting in 1967, the National Hurricane Center has maintained records on the type of cyclogenesis occurring for every Atlantic basin tropical cyclones (eg., Avila 1991). While some terminology has changed since 1967, these summaries are the most complete set of information regarding the interannual variations of the origins of tropical cyclogenesis.

Statistics on tropical storm and hurricane-caused destruction in the United States were obtained from the yearly summaries. Landsea (1991) tabulated this information. Other recent studies (eg., Sheets (1990); Hebert et al. (1992)) normalize long term damage values to current dollars using the Monthly Department of Commerce Implicit Price Deflator for Construction (U.S. Department of Commerce 1991) to account for inflation.

However, to perform a comparative assessment of interannual variations in U.S. damage, the large increases in population along the U.S. coastal zones must also be considered. Sheets (1990) noted a 60% rise in the number of Atlantic coastal inhabitants, a 250% increase in coastal Texas, and an enormous 400% increase in coastal Florida between 1950 and 1985. Obviously, an adjustment only for inflation severely underestimates the potential for damage in areas where populations have increased substantially.

To overcome this deficiency of comparative studies which adjust only for inflation, this paper utilizes both the Implicit Price Deflator for Construction as well as population changes for specific locations. Hurricane King, which struck Miami in 1950 and caused $28 million, provides a good example of the normalization procedure. Since 1950, construction costs have increased by a factor of 5.7 and population along coastal southeast Florida has increased by a factor of 6.0. This combined factor of 34.2 normalizes the damage to $957 million in 1990 dollars.

Intraseasonal Variability

a. Entire Basin Characteristics

Atlantic basin tropical cyclones show a very pronounced seasonal cycle in the level of activity. Neumann et al. (1987) have provided the most recent analysis of the intraseasonal variation of named storms and hurricanes. Atlantic basin tropical cyclone activity either before 1 June or after 30 November is nearly negligible. Both named storms and hurricanes show a strong maximum in mid-September with most cyclones occurring between 1 August and 31 October. This late summer-early fall maximum corresponds to the time of the largest areal extent of warm SSTs and low tropospheric vertical shear in the tropical Atlantic (Gray 1979). However, it is difficult to determine whether named storms and hurricanes have the same peak in terms of percent of annual activity per day.

The annual cycle of IH occurrence in Figure 1 illustrates the "noisy" aspect of the unfiltered data (light curve), reflecting the relatively low climatological incidence of these cyclones. Following Neumann et al. (1987), a nine day running mean was applied to the data, as shown by the solid dark curve. With this filter, the main features of the IH annual cycle appear similar to that identified by Neumann et al. (1987) of the weaker cyclones: namely, a sharp peak of occurrence during mid-September; most activity occurring between 1 August and 31 October; and almost no events before 1 July or after 30 November. The secondary peak in IH activity in late August is likely not physically meaningful.

Figure 2 gives a more quantitative comparison of the annual cycles for various categories of tropical cyclones. This figure shows the daily incidence of tropical storms (maximum sustained wind speed greater than 17 m s-1 but less than 33 m s-1), minor hurricanes (Saffir-Simpson categories 1 and 2), and IH. The total incidence of IH is much less than that of minor hurricanes and tropical storms.

Although this figure indicates that the three categories peak at about the same time, it does not assess the relative strengths of the three maxima. Normalization by percent of annual activity per day accentuates differences in activity among the three groups, as shown in Fig. 3. Intense hurricanes have a much sharper peak during the height of the season. During mid-September, IH experience a maximum of 2.5% of their total annual activity per day, as compared to 1.9% for minor hurricanes and 1.4% for tropical storms. During 1 to 20 September, the average annual occurrence of IH, minor hurricanes, and tropical storms is 43%, 34%, and 25%, respectively. The rankings are reversed for the early and late portions of the hurricane season. In particular, on or before 31 July, the historical mean activity is 13% for tropical storms, 7% for the minor hurricanes, and only 2% for IH. Comparing the variances of the IH with the tropical storm activity, we find that the two distributions are significantly different (at the 0.05 level) using a two-tailed F distribution test (Mendenhall 1979).

The small range of dates during which IH form is likely indicative of the limited duration of favorable environmental conditions available for deep intensification of Atlantic basin tropical cyclones into IH. The more marginal conditions for tropical cyclogenesis (Gray 1979), which occur in this basin from late May to late July and from early November to early December, allow only the occasional tropical storm and infrequent minor hurricane.

b. Spatial Characteristics and U.S. Landfalling Cyclones

For climatological studies of IH which strike the United States, it is instructive to stratify the coastline into two regions. As shown in Fig. 4 (Landsea et al. 1992), these two areas are the Gulf Coast, from Texas to the Florida panhandle, and the East Coast, from the Florida peninsula up to Maine. The entire peninsula of Florida has been included in the East Coast category because cyclones which struck the peninsula were more characteristic, both intraseasonally and interseasonally, of the remainder of the East Coast than the Gulf Coast.

Figure 5 shows that the two geographical regions delineated in Fig. 4 are susceptible to IH strikes at differing times of the year. While the timing of the overall activity is similar to that shown for IH in the entire Atlantic basin in Fig. 1, the Gulf Coast seems to experience IH somewhat earlier in the calendar year than does the East Coast. These differences are explicitly delineated in Table 2 in terms of the first, median, and last date of occurrence.

Figure 6 portrays all IH tracks from 1944 to 1991 by individual months. The regional variations in Figs. 5 and 6 and in Table 2 are likely related to the favorable areas for genesis of tropical cyclones, as well as variations of the general circulation steering. From mid-June to July, IH are infrequent, tend to form in the Gulf of Mexico or in the western Caribbean Sea and generally track toward the west or north. The preferred area of formation and direction of motion favor landfall along the Gulf Coast. From August to early October, intensification of cyclones into IH can occur nearly anywhere in the central and western Atlantic basin between 10N and 35N. But by mid-October, IH form almost exclusively in the western Caribbean or the Atlantic Ocean between the Bahamas and Bermuda and typically begin taking a due north or northeast course, though occasionally moving into the East Coast.

Table 3 summarizes the intraseasonal variability by month. Note that this table suggests that, in general, the amount of August activity is equivalent to that in October. This, at first glance, appears to contradict Table 4 from Neumann et al. (1987), which tabulates the average number of tropical cyclones that form each month. They found that more tropical cyclones form in August than October. However, Table 3 here measures total occurrence, not formation. The different findings for occurrence and formation are reconciled in that while more tropical cyclones form in August than October, more cyclones form in late September and continue into October than form in late July and continue into August.

Interseasonal Variability

a. Atlantic Basin Cyclones

1) Original Best Track Data Set

Year-to-year variations of Atlantic basin IH since 1944 are depicted in Fig. 7. A large range of activity is apparent, varying from no IH occurrences in 1968, 1972, and 1986, to eight major cyclones during 1950. This large year-to-year variability is reflected in a high coefficient of variation (i.e., the ratio of the standard deviation to the mean) of 0.71 which is much larger than that observed for total named storm activity (0.30). This implies that the interannual variability for IH is much greater than weaker cyclones.

As presented in Landsea and Gray (1992), intense hurricane days (IHD) are an additional measure of the seasonal incidence of strong cyclones. One unit of IHD (see Fig. 8) represents four six-hour periods during which an IH was in existence. This type of measure is intrinsically biased to those IH which persist for several days. Historically, long lasting IH originate as strong easterly waves over West Africa and develop in the central North Atlantic. These are termed `Cape Verde' hurricanes (Dunn and Miller 1960), of which Donna-1960, Allen-1980, and Hugo-1989 are notable examples.

Figure 8 details the yearly variations of the Atlantic basin IHD. The seasonal incidence of IHD fluctuates from 0 to 24.5; the coefficient of variation is 0.93 which is even higher than that observed for IH. Both IH and IHD show a substantial decrease in activity with time. This trend may account for a portion of the high coefficients of variation observed.

b. Bias Corrected Best Track Data Set

It is possible that artificial differences of IH activity related to changes in measurement technology are responsible for the apparent trend in the seasonal numbers of IH and IHD. To test for bias between the earlier versus the later decades, the relationship between wind speed and surface pressure was examined. A simplified but useful empirical relationship between the maximum sustained wind speeds and lowest surface pressures of Atlantic basin tropical cyclones was given by Kraft (1961).

V max = (1013-Pc)0.5,

Vmax in this relation is the maximum sustained wind speed in m s-1 and Pc is the minimum sea level pressure in mb.

The Best Track data set contains wind speeds at 6 hourly intervals for the lifetime of each tropical cyclone (even if an estimation was required). The data set does not always include surface pressures which were estimated and archived only from pressure observations and not from the winds (Jarvinen et al. 1984). There have been no significant changes in the measurement of actual or extrapolated surface pressure in tropical cyclones. The methodologies of extrapolating aircraft reconnaissance height data to surface pressures developed in the 1940s and early 1950s (see Jordan 1958) are still being utilized. Hence, we may deduce that any changes in the wind-pressure relationship are due to alterations in the way sustained wind speeds were measured or estimated.

The Best Track data set reports the maximum sustained wind speeds in 5 kt (2.5 m s-1) intervals. A finer tuned estimate of the strength of the cyclones is not feasible with the uncertainties of the various observational platforms. Table 4 presents means and standard deviations of pressure readings stratified for the decades under consideration. For specific wind speed categories, there has been a shift to lower observed pressures, a possible indication that the decades of the 1940s to the 1960s had overestimated wind speeds as compared to later years. In addition, the higher standard deviations for the decades from the 1940s to the 1960s suggest that there was more uncertainty as to what wind speed to assign to strong hurricanes.

To better delineate this bias, Table 5 presents a combined decadal classification which contains at least ten cases per wind division. The corresponding wind speed suggested by Kraft's relationship is included next to the observed minimum pressure. As the criterion used to define IH and IHD in the Best Track data set is a threshold of 100 kt (51 m s-1), it is important to understand the decadal differences (or bias) at that intensity. Hurricanes in the 51 m s-1 category during the years 1970 to 1991 have a mean minimum sea level pressure of 962 mb, almost precisely the pressure required for such a storm by equation (1). However, the mean minimum sea level pressure for the 51 m s-1 hurricane in the 1940s to the 1960s was 968 mb which corresponds to an equivalent wind speed from Kraft of only 48 m s-1. Rather, it is the 54 ms -1 hurricanes in the 1940s to the 1960s that are associated with mean sea level pressure of 963 mb, the highest pressure allowed for IH sta tus by Kraft's relationship. A larger bias of about 5 m s-1 is observed for stronger (Category 4 and 5) hurricanes. However, since the threshold in this study for IH categorization is at 51 m s-1 by the construction of the Best Track data, that higher bias does not affect IH and IHD calculations.

In an attempt to remove the 2.5 m s-1 bias at the IH threshold, the criterion for IH status was increased to 54 m s-1 for the years 1944 to 1969. This effectively reduces the numbers of IH by at most three storms and the numbers of IHD by at most 5 days in this earlier period. The resulting time series of IH and IHD are depicted in Figs. 7 and 8. Table 6 gives the means and standard deviations of the IH activity for the original and adjusted data. Note that the removal of this bias reduces the IH coefficient of variation down to 0.66, still much larger than that observed for named storms.

Removal of the early period bias in IH affects only a portion of the downward trend in IH and IHD. The trend in the bias-removed IH accounts for 13% of the variance in the data (significant at the 0.01 level, from Spiegel (1988)) and explains 8% of the variance in the corrected IHD data (significant at the 0.05 level). Additionally, removal of these trends still leaves a much higher coefficient of variation for IH than named storms (0.62 to 0.30).

However, this reduced activity in recent years might not best be characterized as a linear trend. It appears that IH activity decreased dramatically from 1961 to 1970 and has remained nearly constant through the present time at a much lower frequency than occurred during the 1940s and 1950s. Table 7 contrasts the two long term periods of 1944 to 1960 and 1971 to 1991. The differences in IH and IHD between the time periods are significant at the 0.01 level using the student t test with an adjustment for serial correlation (Reid et al. 1989).

b. U.S. Landfalling Cyclones

The time series of Atlantic basin IH which struck the U.S. coastline between 1899 and 1991 is presented in Fig. 9 (from Landsea et al. 1992). Hebert and Taylor (1978) and Sheets (1990) have previously identified the notable lack of IH affecting the East Coast during the late 1960s to the mid 1980s. This is likely a reflection of the overall basinwide reduction in IH activity during the last two decades as discussed in the previous section. The only other period which incurred a similar (but smaller) reduction in East Coast activity was during the early 1900s to the late 1910s.

Over the same 92 year time frame, the Gulf Coast had nearly the same number of IH as the East Coast but experienced much less multidecadal variation as indicated by the smaller coefficient of variation in Table 8.

The main advantages in using this landfalling cyclone data set are the length of record and consistency of analysis throughout the period (see Hebert and Taylor (1975) for description). No systematic bias would be expected or is found. Additionally, the IH which strike the U.S. mainland can be used, to some degree, as a proxy estimate of variability for IH activity in the entire Atlantic basin. The linear correlation coefficients between landfalling versus total cyclone activity for the years 1944 to 1991 are 0.55 for the original Best Track data set and 0.59 for the bias-removed data set (both significant at the 0.01 level). Thus, while year-to-year variations of landfall activity explain only 35% of the total basin activity, the longer-term decadal activity of the Atlantic basin IH can be approximately represented by the decadal numbers of IH which affect the United States.

c. U.S. Damage

The time series of normalized values for damage (in millions of 1991 dollars) in the United States is presented in Fig. 10. Note the dependence of large damage values on the occurrence of IH. The mean annual damage caused by all tropical cyclones is $1,905 in millions of 1991 dollars. The East Coast suffers more damage on average than the Gulf Coast (59 to 41%, respectively) of the total U.S. damage.

Table 9 presents a summary of the mean and median amounts of damage caused by tropical/subtropical storms and the various Saffir-Simpson categories of hurricanes which affected the United States. Note that median values are more representative of the data set because the means are highly skewed by extremely large outliers. The ``potential damage'' is an idealized normalization of median damage relative to the category 1 hurricanes. This concept illustrates the fact that hurricane damage tends to increase exponentially with intensity. While Hebert and Taylor's (1978) study clearly showed that IH are responsible for most of the U.S. tropical cyclone spawned damage, it is startling to note that category 4 and 5 hurricanes typically do 100 to 250 times the damage of a minimal hurricane. Note that the tropical/subtropical storms are given a `0' for potential damage. Many cyclones striking the U.S. coast with tropical storm or minimal hurricane strength supply enough beneficial rainfall that offsets any minor storm surge or wind-caused damage (Sugg 1968). Of course, occasionally these weaker storms are responsible for major rainfall-induced flooding (eg., Diane in 1955 and Agnes in 1972).

d. Variation of Genesis

An important consideration in understanding the interseasonal variability of the Atlantic basin tropical cyclones is in tracing their origins. The circulation features constituting the largest contributor to tropical cyclogenesis in the Atlantic basin are the West African easterly waves (Avila 1991). On an interannual time scale, Avila showed that the number of easterly waves observed each year is relatively steady; a mean of 60 waves per year with a standard deviation of only 6 is observed. Avila also points out that the number of waves each year has no correlation with the number of Atlantic tropical cyclones. What is noteworthy, however, is the difference in the relative frequency at which easterly waves contribute to the genesis of each of the various tropical cyclone categories. These associations are summarized in Table 10. Note that tropical storms and minor hurricanes (Category 1 and 2 hurricanes) are almost identical (i.e., 57 and 58%) in their relatively moderate rate of easterly wave origins whereas the large majority (83%) of IH form from easterly waves.

Summary and Discussion

Intense hurricanes have substantially different climatological characteristics as compared to the weaker tropical cyclones in the Atlantic basin. On intraseasonal time scales, IH activity experiences a much sharper temporal peak than that of weaker storms with over half of all IH activity occurring during September alone. Interseasonal variability of IH is also much greater than that of tropical storms and minor hurricanes. On multidecadal time scales, basinwide IH activity during the 1970s and 1980s was much reduced from that experienced during the 1940s and 1950s. Note that this reduced level of activity remains apparent in the data after a small bias to overestimation of cyclone intensity during the earlier era is removed. The recent multidecadal pattern of basinwide IH activity as also been reflected in higher amounts of IH strikes in the 1940s and 1950s and a much reduced frequency of IH strikes during the 1970s and 1980s along the U.S. East Coast (Hebert and Taylor (1978) and Sheets (1990)). Also, whereas IH occur less frequently than tropical storms and minor hurricanes, IH are much more dependent upon easterly waves for their formation.

The importance of basic knowledge of IH climatology is underscored by the potential for damage along the U.S. coastal zones which these cyclones can produce. After normalization to 1991 dollars, IH are responsible for an average of $1.3 billion damage per year ($1.9 billion per IH); this is more than 70% of all U.S. tropical cyclone-spawned damage even though they comprise only 20% of all landfalling cyclones. It is likely that a similar ratio also applies in other countries in the Atlantic basin which are affected by IH.

The interannual and interdecadal variability of IH and IHD discussed in sections 4.1 and 4.2 is strongly mirrored by rainfall variations in the Sahel region of West Africa (Gray 1990; Landsea 1991; Landsea and Gray 1992; Landsea et al. 1992). Abundant rainfall in the Western Sahel during the late 1940s through the mid 1960s was concurrent with larger than normal numbers of IH in the Atlantic basin. Conversely, a severe two-decade long Sahel drought occurred with fewer than normal numbers of IH. However, the current Sahel drought is likely a temporary phenomena that will eventually abate (Nicholson 1989; Gray 1990). A return of rainfall in the Sahel, though beneficial for that region, will likely also bring increased IH activity to the Atlantic. The possibility that tropical cyclone-spawned damage will rise dramatically is further accentuated by the large increases of U.S. coastal population during recent decades. Note that these inferences are independent of any potential greenhouse gas warming effects. This climatology is an essential starting point for extrapolations into the future. Further research is needed on the variations of the general circulation which are responsible for the intra- and interseasonal fluctuations that have been described here.


Essential programming assistance has been provided by Richard Taft and William Thorson. Barbara Brumit, Laneigh Walters, and Judy Sorbie-Dunn have contributed help in manuscript and figure drafting preparation. Roger Pielke, John Sheaffer, Pat Fitzpatrick, Ray Zehr, and John Knaff of Colorado State University, Robert Sheets of the National Hurricane Center, Paul Hebert of NWS WSFO in Miami, and Stan Goldenberg of the Hurricane Research Division of NOAA/ERL have made useful comments on an earlier draft of this paper. Additionally, two anonymous reviewers gave very useful critical comments that helped the paper to its final form. And, of course, William Gray, my adviser at Colorado State University has provided invaluable guidance in this work. This research has been supported by NSF and NOAA climate research grants with supplemental support from a NASA Global Change Fellowship.


AMS Council and UCAR Board of Trustees, 1988: The changing atmosphere challenges and opportunities. Bull. Amer. Meteor. Soc., 69, 1434-1440.

Avila, L. A., 1991: Atlantic tropical systems of 1990. Mon. Wea. Rev., 119, 2027-2033.

Clark, G. B., 1960: Hurricane tracking at sea. Mar. Wea. Log, 4, 102-103.

Cry, G. W., 1965: Tropical cyclones of the North Atlantic Ocean. Tech. Paper No. 55, U.S. Weather Bureau, Washington, D.C., 148 pp.

Dunn, G. E. and B. I. Miller, 1960: Atlantic hurricanes. Louisiana State Univ. Press, 326 pp.

Gray, W. M., 1979: Hurricanes: their formation, structure, and likely role in the tropical circulation. Meteorology Over the Tropical Oceans. Roy. Meteor. Soc., James Glaisher House, Grenville Place, Bracknell, Berkshire, RG12 1 BX, 155-218.

Gray, W. M., 1989: Background information for assessment of expected Atlantic hurricane activity for 1989. 11th Annual National Hurricane Conference, Miami, FL, 41 pp.

Gray, W. M., 1990: Strong association between West African rainfall and U.S. landfall of intense hurricanes. Science, 249, 1251-1256.

Hebert, P. J. and R. A. Case, 1990: The deadliest, costliest, and most intense United States hurricanes of this century (and other frequently requested hurricane facts). NOAA Tech. Memo. NWS NHC 31, Miami, Florida, 31 pp.

Hebert, P. J., J. D. Jarrell, and M. Mayfield, 1992: The deadliest, costliest, and most intense hurricanes of this century (and other frequency requested hurricane facts). NOAA Tech. Memo. NWS NHC 31, Miami, FL, 39 pp.

Hebert, P. J. and G. Taylor, 1975: Hurricane experience levels of coastal county populations, Texas to Maine. Special Report, National Weather Service Community Preparedness Staff and Southern Region, 153 pp.

Hebert, P. J. and G. Taylor, 1978: The deadliest, costliest, and most intense United States hurricanes of this century (and other frequently requested hurricane facts). NOAA Tech. Memo. NWS NHC 7, Miami, Florida, 23 pp.

IPCC, 1990: Climate change: The IPCC scientific assessment. Rept. prepared by Working Group I, J. T. Houghton, G. J. Jenkins, and J. J. Ephramus, Eds., Cambridge Univ. Press.

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. NOAA Tech. Memo. NWS NHC 22, Miami, Florida, 21 pp.

Jordan, C. L., 1958: Estimation of surface central pressure in tropical cyclones from aircraft observations. Bull. Amer. Meteor. Soc., 39, 345-352.

Kraft, R. H., 1961: The hurricane's central pressure and highest wind. Mar. Wea. Log, 5, 155.

Landsea, C. W., 1991: West African monsoonal rainfall and intense hurricane associations. Dept. of Atmos. Sci. Paper No. 484, Colo. State Univ., Fort Collins, Colorado, 272 pp.

Landsea, C. W., and W. M. Gray, 1992: The strong association between western Sahel monsoon rainfall and intense Atlantic hurricanes. J. Climate, 5, 435-453.

Landsea, C. W., W. M. Gray, P. W. Mielke, Jr., and K. J. Berry, 1992: Longterm variations of western Sahel monsoon rainfall and intense U.S. landfalling hurricanes. J. Climate, 5, 1528-1534.

McAdie, C. J. and E. N. Rappaport, 1991: Diagnostic report of the National Hurricane Center, 4, No. 2, National Hurricane Center, Coral Gables, Florida, 75 pp.

Mendenhall, W., 1979: Introduction to probability and statistics. Duxbury Press, North Scituate, Massachusetts, 594 pp.

Mitchell, C. L., 1924: West Indian hurricanes and other tropical cyclones of the North Atlantic Ocean. Mon. Wea. Rev. Supplement No. 24, U.S. Weather Bureau, Washington, D.C., 1924, 47 pp.

Neumann, C. J., B. R. Jarvinen, A. C. Pike, and J. D. Elms, 1987: Tropical Cyclones of the North Atlantic Ocean, 1871-1986. National Climatic Data Center in cooperation with the National Hurricane Center, Coral Gables, Florida, 186 pp.

Nicholson, S. E., 1989: Long-term changes in African rainfall. Weather, 44, 46-56.

Reid, G.C., K.S. Gage, and J. R. McAfee, 1989: The thermal response of the tropical atmosphere to variations in equatorial Pacific sea surface temperature. J. Geophys. Res., 94, 14705-14716.

Sheets, R. C., 1990: The National Hurricane Center - past, present, and future. Wea. Forecast., 5, 185-232.

Simpson, R. H, 1974: The hurricane disaster potential scale. Weatherwise, 27, 169-186.

Spiegel, M. R., 1988: Statistics - Schaums Outline Series. 2d ed., McGraw- Hill, 504 pp.

Sugg, A. L., 1968: Beneficial aspects of the tropical cyclone. J. Appl. Meteor., 7, 39-45.

Sumner, H. C., 1944: North Atlantic hurricanes and tropical disturbances of 1944. Mon. Wea. Rev., 72, 237-240.

U.S. Department of Commerce, 1991: Costs, prices, and interest rates. Construction Review, 37, 38.