In a challenge to the conclusions of that editorial, Mike MacCracken - former Director of the U.S. Global Change Research Program and Executive Director of the U.S. National Assessment Coordination Office - wrote to climatesceptics@yahoogroups.com claiming "it is generally agreed that there is no evidence to suggest a trend over the 20th century in Atlantic hurricanes."  Why?  Because the Intergovernmental Panel on Climate Change (IPCC) "says this," he says; and in support of this claim, he notes that the IPCC cites the paper of Fernandez-Partagas and Diaz (1996) to this effect.  Consequently - and in spite of the fact that the two pertinent hurricane files of our Subject Index referenced in the first paragraph of this editorial present synopses of twelve peer-reviewed scientific journal articles based on real-world data that document the existence of declining trends in the frequency and/or intensity of Atlantic hurricanes during periods of rising global air temperature - MacCracken suggests that "we all need to go back to agreeing that there is no clear trend in hurricane frequency or intensity over the 20th century."

Since it seems a bit of a stretch to us to set aside the conclusions of a full dozen scientific studies on the basis of a single conflicting study, we thought it important to take a closer look at the Fernandez-Partagas and Diaz paper, as well as what its authors have to say about their findings.

This study, in the words of its authors, consisted of "a historical revision of Atlantic tropical cyclones for the period 1851-90," wherein they identified "many tropical storms and hurricanes that had not been previously documented."  This feat was accomplished by meticulously searching a number of historical records for evidences of storms that had been missed by earlier studies.

For the entire 40-year period, the authors' Herculean efforts boosted the previously-published total number of such storms by 55%; while for the first 30 years of the period it enhanced the prior count by 76%, and for the first 20 years it increased the original tally by close to 150%.  Nevertheless, when the upward-revised 1851-90 total count of Atlantic hurricanes was compared to the 1951-90 total count, it was found that the earlier cooler period still registered approximately 12% fewer tropical storms than the more recent warmer period, which is indeed suggestive of an increase in storm frequency in response to the documented global warming that occurred between 1851 and 1990, which result is contrary to the implications of the findings of the several studies of this subject that we have reviewed on our website.

The question we must ask ourselves, however, is how accurate are the revised storm numbers for the first 40-year period?  Fernandez-Partagas and Diaz demonstrated that earlier studies had missed a massive amount of hurricanes during this period.  Is it not logical, therefore, to assume that they too might have missed a lot of them?

The authors themselves address this important question in discussing their findings, acknowledging that even with the large upward adjustments they made in the nineteenth-century storm numbers, "the totals in the nineteenth century are still likely to be a significant underestimate of the actual storm totals for this time, due to the much poorer system of communication and sparser population of that time, as well as to the obvious lack of modern observational tools, such as satellite and aircraft reconnaissance."

So just how large is this significant underestimate?  No one can say for sure; but a mere 12% increase in the early-period count would equalize the storm numbers of the pertinent 19th- and 20th-century 40-year intervals.  Considering the fact that their first cut at this problem boosted the original 1851-90 storm tally by 55%, a 12% increase - or even something larger - does not seem unreasonable, especially when recalling that the authors' detective work enhanced the storm count of the 1850s by nearly 150%, yet they still volunteered their opinion that the resultant number of storms in the 1850s was so low as to suggest that it "may not be real."

In light of these considerations, which were largely advanced by Fernandez-Partagas and Diaz themselves, their admittedly-biased study would appear to provide little reason for rejecting what the many studies we have cited clearly demonstrate, i.e., that increases in mean global air temperature - whether natural or anthropogenic-induced, depending on one's point of view - have a tendency to reduce both the frequency and intensity of Atlantic basin hurricanes.  Since this topic is of such great importance, however, we have searched the literature for still other pertinent papers that may be relevant to the issue.

Perhaps the most important study we found - which might at first be construed to contradict what we have said about the subject - is that of Goldenberg et al. (2001), who note that relative to the 24-year period 1971-1994, the six-year period 1995-2000 saw a doubling of overall hurricane activity in the North Atlantic and a 2.5-fold increase in major hurricanes there.  Here too, however, the authors say it is possible that "the current active period (1995-2000) only appears more active than the previous active period (1926-1970) due to the better observational network now in place," additionally stating that "further study is essential [our italics] to separate any actual increase from an apparent one due to more complete observations" being made in the more recent period

In addition to this obvious truth, Goldenberg et al. make a number of other interesting observations.  They note, for example, that studies of global sea surface temperatures (SSTs) "have shown that the primary source of interannual SST variability is the ENSO region."  In this regard - and in harmony with the conclusions of the several studies of El Niņo effects on Atlantic hurricanes described in our Subject Index - they report that in the midst of the 1995-2000 period of significantly heightened Atlantic hurricane activity, there was one year (1997) that still exhibited "below average activity," which occurred, as they note, "when the Atlantic hurricane activity was suppressed by the strongest El Niņo event of this century."

To remove such large-scale El Niņo-induced global warming effects and thereby enhance their ability to detect and evaluate more localized climatological impacts on Atlantic hurricane characteristics, Goldenberg et al. removed the teleconnected effects of ENSO on the Atlantic Ocean.  When this feat was accomplished, they were left with what they called the "Atlantic miltidecadal mode," the positive phase of which "has warm SSTs in the tropical North Atlantic from 0° to 30°N and in the far North Atlantic from 40° to 70°N."

With respect to this phenomenon, Goldenberg et al. observed that "the time series for the Atlantic multidecadal mode, major hurricanes and NTC [net tropical cyclone activity] all show similar [i.e., in-phase] multidecadal-scale shifts."  What is more, in their plot of the history of the Atlantic multidecadal mode back to 1870, it can be seen that this important determinant of Atlantic basin hurricane activity exhibited even stronger positive values during much of the previous period of heightened Atlantic hurricane activity (1926-1970) than the high values that have occurred during the current active period (1995-2000), which suggests that the hurricane activity of the current active period is likely not any greater - and could easily be even less - than that of the earlier active period, as Goldenberg et al. have indeed suggested may well be the case.

In another interesting paper - which is an update of the Fernandez-Partagas and Diaz (1996) study and is based on the results of an even more recent hurricane reanalysis project - Elsner and Bossak (2001) looked at the same 1851-1990 time period, but with a ten-year extension to the year 2000 that includes the most recent six-year period of intense hurricane activity studied by Goldenberg et al.  For the first two-thirds of this interval, i.e., from 1851-1950, the mean annual number of all U.S. landfalling hurricanes was determined to be 1.80, while for the 1951-2000 period it was found to be only 1.44, or 20% lower.  Likewise, corresponding mean annual numbers for major U.S. landfalling hurricanes were found to be 0.62 and 0.54, indicative of a reduction of 13% in the latter 50-year period, even though the last six years of this interval exhibited a pronounced increase in hurricane activity.  Hence, both of these results bear out the suspicions of Fernandez-Partagas and Diaz that their hurricane counts for the last half of the nineteenth century were indeed much too low.

Interestingly, Elsner and Bossak report that even their new-and-improved hurricane counts for the last half of the nineteenth century may be too low.  They note, for example, that "the variance of interannual activity decreases from 2.3 during the last half of the nineteenth century to 1.6 during the last half of the twentieth century, indicating a potential bias during the earliest 50-year period."  With respect to this problem, they say that the "larger interannual variance (lower precision) during the nineteenth century might result from an incomplete record," because, as they rightly note, "a hurricane striking southeastern Florida or southern Texas during the 1850s could have gone undetected as these areas were undeveloped at that time."

That this problem does indeed still plague the first third of Elsner and Bossak's long-term hurricane record is suggested by the fact that their all-hurricane and major-hurricane counts for this period are still 4% and 28% lower than their corresponding counts of the following 50-year period.  Consequently, any additional improvements that might be made to their Atlantic hurricane histories, as well as to those of some of the other histories listed in our Subject Index, would be expected to significantly raise their early-period hurricane numbers and thereby amplify the long-term historical declines in all-hurricane and intense-hurricane numbers that these studies have documented.

In addition to the studies of Goldenberg et al. (2001) and Elsner and Bossak (2001), which pertain to the behavior of Atlantic hurricanes over the past 150 years of global warming (Esper et al., 2002), we found some additional papers that confirm our earlier observations about El Niņo conditions leading to fewer Atlantic basin hurricanes than neutral or La Niņa conditions:

• Bove et al. (1998) - "The probability of two or more landfalling U.S. hurricanes during an El Niņo is 28%, of two or more landfalls during neutral conditions is 48%, and of two or more landfalls during La Niņa is 66%... The probability of one or more major hurricane landfalls during El Niņo is 23% but is 58% during neutral conditions and 63% during La Niņa."

• Kimberlain and Elsner (1998) - "El Niņo conditions ? are linked to hostile environmental conditions for tropical cyclogenesis and hurricane formation over the tropical belts of the North Atlantic."

• Elsner et al. (1999) - "Clearly the cold phase is more conducive to U.S. hurricane activity.  Moreover, the average annual number of U.S. tropical-only hurricanes during the cold phase of ENSO is 1.5 compared with 0.2 during the warm phase."

Even climate modelers are getting into the act now that data-based studies have blazed a well-marked trail in this area.  As reported by Vitart and Anderson (2001), for example, using a ten-member ensemble of atmospheric general circulation model simulations forced by climatological SSTs of the 1980s everywhere but over the tropical Pacific and Indian Oceans, they have demonstrated that "significantly fewer [Atlantic] tropical storms are simulated with El Niņo SSTs imposed over the tropical Pacific and Indian Oceans than with La Niņa condtions."

Beyond this point, further discussion of MacCracken's response to our Editorial of 24 April 2002 would be rather meaningless, as he bases of all his contentions on what he calls the "premise" that Atlantic hurricane frequency and intensity do not decline with increasing global air temperature, which premise is clearly not supported by the great bulk of modern research data.  When he asks in the second of his numbered comments, for example, "why is there not a downward trend in hurricanes during the 20th century," we simply say there is!  And when he asks "why is the number of hurricanes level," we say it isn't!

Like a broken record, MacCracken maintains this same approach to the issue in the third of his numbered challenges of our position, asking "why was there not a downward trend in the number of Atlantic hurricanes?"  Likewise, he begins his fourth complaint by saying "to explain the constancy in the number of hurricanes with the global warming..."  His usage of this faulty premise undercuts the validity of nearly everything he has to say.

MacCracken's fifth and final point, however, stretches our credulity in an entirely new direction; for he uses computer-model results to argue against the veracity of real-world observations, in this case, against the observation that decreases in Atlantic hurricane intensities, as well as the numbers of intense Atlantic hurricanes, tend to occur during periods of rising mean global air temperature.  We would hope it would be readily obvious to most people that when model predictions disagree with what is observed, it is the models that are wrong, not nature.

Clearly, real-world observations trump everything, even the "premises" of one of the United States' foremost climate officials, as well as those of any other individuals or organizations that might be tempted to turn a blind eye to the world around us.  In science, there is no authority, only data!

Reference
Bove, M.C., Elsner, J.B., Landsea, C.W., Niu, X. and O'Brien, J.J.  1998.  Effect of El Niņo on U.S. landfalling hurricanes, revisited.  Bulletin of the American Meteorological Society 79: 2477-2482.

Elsner, J.B. and Bossak, B.H.  2001.  Bayesian analysis of U.S. hurricane climate.  Journal of Climate 14: 4341-4350.

Elsner, J.B., Kara, A.B. and Owens, M.A.  1999.  Fluctuations in North Atlantic hurricane frequency.  Journal of Climate 12: 427-437.

Esper, J., Cook, E.R. and Schweingruber, F.H.  2002.  Low-frequency signals in long tree-ring chronologies for reconstructing past temperature variability.  Science 295: 2250-2253.

Fernandez-Partagas, J. and Diaz, H.F.  1996.  Atlantic hurricanes in the second half of the nineteenth century.  Bulletin of the American Meteorological Society 77: 2899-2906.

Goldenberg, S.B., Landsea, C.W., Mestas-Nuņez, A.M. and Gray, W.M.  2001.  The recent increase in Atlantic hurricane activity: causes and implications.  Science 293: 474-479.

Kimberlain, T.B. and Elsner, J.B.  1998.  The 1995 and 1996 North Atlantic hurricane seasons: a return of the tropical-only hurricane.  Journal of Climate 11: 2062-2069.

Vitart, F. and Anderson, J.L.  2001.  Sensitivity of Atlantic tropical storm frequency to ENSO and interdecadal variability of SSTs in an ensemble of AGCM integrations.  Journal of Climate 14: 533-545.