William M. Gray
Department of Atmospheric Science
Colorado State University
Fort Collins, Colorado
Paul W. Mielke
Department of Statistics
Colorado State University
Fort Collins, Colorado
Kenneth J. Berry
Department of Statistics
Colorado State University
Fort Collins, Colorado
Journal of Climate Vol. 5,1528-1534 (1992)
AMS Copyright Notice
© Copyright 1992 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 firstname.lastname@example.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.
This paper extends our analysis of the African rainfall and Atlantic hurricane association back to 1899 utilizing the best available precipitation and tropical cyclone data. The additional 50 years of data appear to confirm our earlier assessme nts; we find that the Sahelian rainfall-hurricane relationship is highly significant for the Florida peninsula and the remaining U.S. East Coast states, both together and separately, and that no association can be shown for the U.S. Gulf Coast.
Nevertheless, reliable tropical cyclone data are available for hurricanes that made landfall along the U.S. mainland back to 1899. The large coastal populations in these areas make it unlikely that any hurricanes passed unnoticed across the U.S. coast sin ce the turn of the century. A dataset by Neumann et al. (1987) categorizes these hurricanes by intensity. The Saffir-Simpson Scale (Simpson 1974), shown in Table 1, is used to rank the hurricanes from a 1 (minimal) to a 5 ( catastrophic). Saffir-Simpson Category 3, 4, and 5 hurricanes are collectively referred to as "intense," while Categories 1 and 2 are characterized as "minor."
Not every hurricane that forms in the Atlantic basin makes landfall in the United States; however, the eastern U.S. coastline is so extensive that it often lies directly in the path of these cyclones. Just less than two of the basin's annual av erage of six hurricanes hit the U.S. coast. Consequently, more hurricanes strike the United States during relatively active hurricane seasons, while the United States usually experiences fewer landfalling hurricanes in calm seasons. The linear correlation between seasonal U.S. landfalling hurricanes and total seasonal Atlantic hurricanes is r = 0.50 for the years 1949 to 1990. Therefore, while not corresponding precisely on a year-to-year basis, the long-term variations of U.S. landfalling hurricanes back to 1899 provide a good estimate of the overall variations of the total Atlantic basin activity for decades prior to aircraft reconnaissance.
For the purpose of characterizing hurricane activity, the eastern U.S. coastline can be approximately separated into three regions: the Gulf Coast (including panhandle Florida), the Florida peninsula, and the upper Atlantic Coast as seen in Fig. 1. The Florida peninsula and the upper Atlantic Coast are collectively referred to as the East Coast, as defined in Landsea and Gray (1991a). In constructing the time series of U.S. landfalling hurricanes, cyclones that made two separate landfalls along the United States are counted twice (for example, Hurricane Betsy in 1965 struck south Florida, reintensified over the Gulf of Mexico, and then hit Louisiana). This procedure is appropriate in view of the immense destructive potential of a hurricane (such as Betsy) that makes two distinct and separate landfalls. (Nevertheless, counting such storms singly results in almost identical statistical conclusions.)
While the Gulf Coast intense hurricanes have experienced a relatively constant frequency, East Coast intense hurricanes have decreased sharply during the last 25 years as shown in Fig.2. Most of this East Coast reduction is due t o the extreme lack of intense hurricane strikes along the Florida peninsula and, to some degree, a lessening of strikes along the upper Atlantic Coast. During this century, a reduction of hurricane activity of this magnitude is without precedent, with onl y the period from 1900 to 1918 vaguely reflecting the decrease seen recently. However, as pointed out by Sheets (1990), this diminished activity in recent years, while being beneficial in the short term, may lead to disastrous consequences "when hurricane activity inevitably returns to the frequencies experienced during the 1940s-60s." Likewise, "the decreased [hurricane] death total [in recent years] may be as much a result of [a] lack of major hurricanes striking the most vulnerable areas as they are of any fail-proof forecasting, warning, and observing system" (Hebert and Case 1990). Thus, it is necessary to identify associated meteorological features so that a better understanding of these multidecadal variations can be attained and adequate preparati on taken when the frequency of intense hurricanes that affect the United States increases.
Our previous work with West African data utilized a 38-station index for the June to September western Sahelian rainfall (Gray 1990; Landsea and Gray 1991a). For each year, individual station rainfall for this 4-month period is compared against the statio n's long-term mean and standard deviation to define individual station deviation values (i.e., a z score). The regional index value for that year is simply the average of the available station deviations. This type of index formulation, which avoids biasi ng the data toward either normally dry or wet stations, was first described by Kraus (1977).
Unfortunately, only five stations in the western Sahel are still operating that have relatively continuous records back to the beginning of this century (Fig. 3). These five stations are used to form our "turn of the century" (TO C) western Sahelian rainfall index; the time series of this index is shown in Fig. 4. Table 2 is a tabular listing of the time series ranked from wettest to driest and Tabl e 3 provides statistics for the individual stations.
Note that while the available data show a wide range of rainfall means (Table 3) for the 4-month period, the individual stations correlate at a level of at least r = 0.59 versus the combined TOC index. In addition, this five-station index shares 83% of its variance with the larger 38-station western Sahelian index (Landsea and Gray 1991a) when compared for the common years 1949 to 1990. Thus, the TOC rainfall index is representative of individual locations in the western Sahel and is compatible with the more dense network of stations available in recent years.
Strongly apparent in the time series of the rainfall in Fig. 4 is the recent and unprecedented dry period that appears to be continuing at the present time. Aside from a few years in the periods between 1899 to 1914 and 1937 to 1 942, the rainfall was consistently more abundant than the amounts which have been reported since the late 1960s. These features, which qualitatively appear similar to the intense hurricane activity (especially along the U.S. East Coast), will be shown to be quantitatively significant as well.
The Sahelian rainfall data are ranked from the wettest (1906 rank 1) to the driest year (1983 rank 92) in Table 2. The recent extreme drought is represented in this ranking in that 18 of the 23 driest years occurred between 1968 and 1990. The wettest 23 years occur throughout a wider range of decades, with 3 to 5 wet years observed from each decade between 1900 and 1959.
With a ranked dataset 92 years in length, one can test the significance levels of the hurricane-Sahelian rainfall associations by making comparisons of rainfall quartiles that contain 23 years of data in each. Table 4 pr esents data for U.S. landfalling intense hurricanes and minor hurricanes, stratified by the amount of western Sahelian rainfall. While the minor hurricanes show no relationship with western Sahelian rainfall, the East Coast intense hurricanes show extreme differences both in the wet-half versus the dry-half stratification and in the quartile analysis. Strong significance is also realized in the further stratification of the East Coast into the two subregions: the Florida peninsula and the upper Atlantic Coast. Note that even with the addition of the Gulf Coast (which shows no association), more than three times as many intense hurricanes are seen along the entire U.S. coast during the 23 wettest years as compa red to the 23 driest years. Correlation coefficients for western Sahelian rainfall versus intense hurricane strikes are r = 0.11 (not significant) for the Gulf Coast, r = 0.23 (significant at the 0.05 level) for the Florida peninsula, and r = 0.32 (signif icant at the 0.01 level) for the upper Atlantic Coast.
Figure 5 contrasts the tracks of U.S. landfalling intense hurricanes during the wettest and driest quartiles. This figure graphically illustrates that the large wet-to-dry differences are manifested primarily along the East Coast (the Florida peninsula and the upper Atlantic Coast). Along the Gulf Coast, the quartile ratio in intense hurricanes is only 10 to 7 fo r the wettest versus the driest years. However, for the area extending from the Florida peninsula up to Maine, the East Coast, the ratio is an extreme 15 to 0. An additional feature of Table 4 indicates that the East Coast variations arise from nearly equal contributions of intense hurricanes striking the Florida peninsula and the upper Atlantic Coast. Both subregions received int ense hurricane strikes about once every three years during the wettest 23 western Sahelian years, whereas none occurred during the 23 driest years. As discussed in Landsea (1991), of the various types of Atlantic tropical cyclones, the intense hurricanes are of most interest. The large downward trend observed in the incidence of these cyclones striking the United States is also observed in the freque ncy of intense hurricanes affecting the Caribbean islands, as well as overall intense hurricane activity. Along the U.S. coast intense hurricanes are responsible for over three-quarters of the tropical cyclone-spawned damage, even though on average they s trike only twice every three years.
Variations of U.S. landfalling intense hurricane activity are shown to be strongly related to western Sahelian rainfall during the last 92 years. This association is strongest for the U.S. East Coast and is significant both for the Florida peninsula and t he upper Atlantic Coast. Some physical mechanisms likely to account for such strong associations are detailed in Gray (1990) and Landsea and Gray (1991a); these include drought versus rainy variations in the amount of tropospheric vertical wind shear over the Atlantic basin and in the strength of African-spawned easterly waves.
An additional possibility is rainfall-related variations in the tropical cyclone steering patterns. Analogous to early work by Namias (1955) and Ballenzweig (1959), the frequency with which hurricanes strike the United States is highly dependent upon trop ospheric flow patterns. It is likely that a lack of U.S. intense hurricanes and drought in the western Sahel are related by the presence of a stronger-than-normal central Atlantic midtropospheric trough, which would act to recurve any tropical cyclones aw ay from the U.S. coast. The 1990 hurricane season (Mayfield and Lawrence 1991) may be an example of both unfavorable vertical shear and a stronger central Atlantic trough being responsible for reduced Atlantic basin intense hurricane activity and no U.S. landfalling intense hurricanes during a western Sahelian drought year.
An interesting aspect of the Sahelian rainfall-hurricane association is that a strong relationship is not observed for intense hurricanes striking the Gulf Coast. It is likely that the Gulf of Mexico may be considered distinct subbasin, and that anomalous general circulation features associated with Sahelian drought and moist years are subordinate to influences that are specific to the Gulf of Mexico. It is known that intense hurricanes in the Gulf of Mexico have characteristics that are different from th e remainder of the Atlantic basin (Landsea 1991). Specifically, these cyclones typically occur almost three weeks earlier in the season on average (median date of landfall along the Gulf Coast is 5 September) and their origins are often from midlatitude s ystems. Since 1967 (when satellite monitoring made it possible), only intense hurricanes that were spawned from African waves have made landfall along the East Coast, while midlatitude systems (e.g., stationary frontal boundaries or upper-tropospheric cut off lows) can occasionally form an intense hurricane that makes landfall along the Gulf Coast. Hurricane Alicia, which struck the Texas coast in 1983, is a notable example of this latter phenomenon.
When the western Sahel obtains a reprieve from its multidecadal drought with a return of significant rainfall, it is very probable that the Atlantic basin, including the U.S. coastal regions and the Caribbean islands, will experience more intense hurrican e activity. These hurricanes are of most concern because of their potential destructiveness and because the recent decrease of activity may have led to complacency and a lack of community preparedness. Nicholson (1989) has shown that previous droughts in the Sahel have been replaced by episodes of abundant rainfall. Hence, there is currently no reason to suspect that the processes responsible for the drought affecting the region and contributing to the lack of intense hurricanes will continue indefinitely .
The original presentation of the concepts described in this paper was made at the AMS 5th Climate Variations Conference (Landsea and Gray 1991b). The authors are grateful for all of the West African rainfall data that have been supplied to us by W. Spangl er and R. Jenne at NCAR ( due in most part to the extensive collection by S. Nicholson at FSU), G. Farmer of U.S. A.I.D. FEWS Project, P. J. Lamb of the University of Oklahoma, D. Miskus and R. J. Tinker of CAC, and E. O. Oladipo of Ahmodu Bello Universit y of Nigeria. Excellent programming assistance has been provided by R. Taft and W. Thorson in the processing of the data. B. Brumit, L. Walters, and J. Sorbie-Dunn have provided important manuscript, data analysis, and drafting assistance. John Sheaffer o ffered valuable manuscript editorials. This research has been funded by the National Science Foundation Grant ATM-8920645.
Gray, W. M., 1990: Strong association between West African rainfall and U.S. Iandfall of intense hurricanes. Science, 2A9, 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 facts). NOAA Tech. Memo., NWS NHC 31, 31 pp.
Kraus, E. B., 1977: Subtropical droughts and cross equatorial transports. Mon. Wea. Rev., 105, 1009-1018. c
Landsea, C. W., I 991: West African monsoonal rainfall and intense hurricane associations. M.S. thesis, Dept. of Atmospheric Science, Colorado State University, Paper No. 484, 272 pp.
Landsea, and W. M. Gray, 1991a: The strong associations between Western Sahel monsoon rainfall and intense Atlantic hurricanes. J. Climate, 435-453.
Landsea, and W. M. Gray, 1991b: Concurrent multidecadel variations of West African monsoonal rainfall and intense hurricane activity. Preprints, Fifth Climate Variations Conference, Denver, Amer. Meteor. Soc., 38-41.
Mayfield, M., and M. B. Lawrence, 1991: Atlantic hurricane season of 1990. Mon. Wea. Rev., 119, 2014-2026.
Namias, J., 1955: Secular fluctuations in vulnerability to tropical cyclones in and off New England. Mon. Wea. Rev., 83, 155162.
Neumann C. J., B. R. Jarvinen, A. C. Pike, and J. D. Elrns, 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.
Rosenthal, S. L., 1991: A note on relationships between Western Sahel rainfall and U.S. hurricane activity. NOAA Tech. Memo., ERL AOML-68, 18 pp.
Sheets, R. C., 1990: The National Hurricane Center Past, present, and future. Wea. Forecasting, 5, 185-232.
Simpson, R. H., 1974: The hurricane disaster potential scale. Weatherwise, 27, 169-186.