Reply to "Comment on Changes in the Rates of North Atlantic Major Hurricane Activity During the 20th Century"

James B. Elsner

In Elsner et al. [2000] (hereafter EJN00) we provide an elegant statistical analysis of North Atlantic hurricane data that reveals temporal changes to major hurricane activity during the 20th century. For the most part results provide information that is common knowledge, but the procedure is rooted in firm scientific reasoning. Landsea [2001] (hereafter L01) comments on this work. Unfortunately the remarks are either wrong or deceptive, misleading the reader about the nature and significance of our research. The comments L01 make are threefold: 1) he claims the treatment of the database is erroneous, 2) he claims the proposed link to the North Atlantic oscillation (NAO) is unsubstantiated, and 3) he claims there is a neglect of key earlier studies. Responses to these allegations are provided below.

Regarding the first claim of L01, in EJN00 we state ``The change-point model detects shifts in the rates of occurrence regardless of their physical or non-physical origins. The change at 1943 is due in part to improvement in our capabilities to observe these storms." Thus I believe it is deceptive of L01 to allege that we mistakenly treated the entire record as trustworthy. Our analysis correctly confirmed the observed rate change in 1943. The model detected additional rate changes around 1965 and 1995 that are likely real and which are not a result of observational bias. We kept the earlier unreliable part of the record for two reasons: 1) to verify the model's capability to detect a likely change point around 1943 and 2) to demonstrate that the hurricane record from the earlier biased period shows a remarkably similar rate to the record from the later unbiased period.

L01 advances the notion that the set of all hurricanes is biased downward before 1943. We tend to agree, but to what extent is open to considerable debate (see Elsner and Kara [1999]). L01 shows that the correlation between all major hurricanes to those hitting the coast is different before and after this year, but he offers no formal test of statistical significance. Moreover, the statement about the work of Solow and Moore [2000] is incorrect. The results of Solow and Moore [2000] show a significant reduction in Atlantic basin-wide hurricane activity over the period 1930-98. This is different than the results given in L01. What is similar is that both L01 and Solow and Moore [2000] use the same-rather coarse- assumption that the proportion of basin-wide hurricanes that make landfall is constant over time. Neither L01 nor Solow and Moore [2000] test this assumption.

Regarding the second claim of L01, we show in EJN00 that a Poisson regression of annual major hurricane counts against Iceland sea-level pressures provides a statistically significant model with a p-value of 0.023. The p-value arises from the probability of the c2 value being less than the deviance difference with a single degree of freedom. Therefore L01 is mistaken; careful readers are not left wondering about the statistical significance of this relationship.

Finally, in EJN00 we do not claim to have ``discovered" multidecadal variability in the major hurricane record. EJN00 cite the peer-reviewed work of Wilson [1997, 1999] and Landsea et al. [1996] on this subject. Thus, I believe L01 is incorrect to contend that we neglected previous work on this topic, and to use the adjective ``deliberate" is laughable. The importance of EJN00 is that we go beyond the published literature on this subject by putting the debate about the significance of 1995 on firm statistical footing. Speculation on potential climate shifts at conferences and in non-peer reviewed literature, and the use of limited statistical tools are no substitute for proper quantitative studies.

EJN00 provide a correct, peer-reviewed, and refined quantitative analysis that detects observed de-cadal variations in North Atlantic major hurricane activity over the past century. The work is important in placing widespread but diffuse speculation concerning recent changes to North Atlantic hurricane activity on solid statistical grounds.

The comments of S. Elsner are appreciated.

Elsner, J. B., and A. B. Kara, Hurricanes of the North Atlantic: Climate and Society, Oxford, New York, 488 pp, 1999.

Elsner J. B., T. Jagger, and X.-F. Niu, Changes in the rates of North Atlantic major hurricane activity during the 20th century, Geophys. Res. Lett., 27, 1743-1746, 2000.

Landsea, C.W., Comment on ``Changes in the rates of North Atlantic major hurricane activity during the 20th century," Geophys. Res. Lett., 2001.

Landsea, C. W., N. Nicholls, W. M. Gray, and L. A. Avila, Downward trends in the frequency of intense Atlantic hurricanes during the past five decades, Geophys. Res. Lett., 23, 1697-1700, 1996.

Solow, A. R., and L. Moore, Testing for a trend in a partially incomplete hurricane record, J. Climate, 13, 3696-3699, 2000.

Wilson, R. M., Comment on ``Downward trends in the frequency of intense Atlantic hurricanes during the past 5 decades," by C.W. Landsea et al. Geophys. Res. Lett., 24, 2203-2204, 1997.

--, Statistical aspects of major (intense) hurricanes in the Atlantic basin during the past 49 hurricane seasons (1950-1998): Implications for the current season, Geophys. Res. Lett., 26, 2957-2960, 1999.

File translated from TEX by TTH, version 2.91.
On 18 May 2001, 16:06.