Predictability of hurricane activity and impacts

Human and economic casualties represent the convolution of the geophysical or climatological threat with property or populations at hazard. It is intriguing that US damage normalized for inflation, population and individual wealth exhibits zero trend from 1900 through 2008 (Pielke et al. 2008). Analysis of stationary time series is a powerful tool in geophysical science, and the power spectrum of normalized damage provides insight into the short-term climatology of the hurricane threat.

Here the US damage time series is extended to 128 years by adding damage from analog storms during the years 1881-1899. Each 1881-1900 landfall was identified with one of five 1900-2008 landfalls for which the normalized damage is known by calculating the “distance” between the historical and analog landfall, S² = D² + [(V₁₉ – V_a)/20]². D is the great circle separation of the landfall points in degrees of latitude, V₁₉ is the landfall maximum wind of the 19th Century storm, and V_a is that of the each analog storm. The term in square brackets is essentially the difference in intensities expressed as Saffir-Simpson categories so that a geographical separation of a degree increases S2 as much as one category difference in intensity.

The closest analog landfalls are those with the five smallest values of S². The spectra of the original data padded with zeros during 1881-1900 and five spectra padded with randomly selected closest analog storms were all essentially random for frequencies from 1/128 yr^-1 to the Nyquist Frequency, 1/2 yr^-1. Apart from unimpressive (and not significant) peaks identifiable with the 2-to-5-year ENSO cycle and an intriguing (also not significant) peak that lines up with the 11-year sunspot cycle (e.g. Elsner et al. 1999), the spectrum in essentially white noise. It is exactly what one would expect for a time series where much of the variance arises from a few large-amplitude impulses, and it implies essentially no predictability on any timescale shorter than a century.

When the effect of the Atlantic Multidecadal Oscillation (AMO) on hurricanes was first described (Goldenberg et al. 2001), the investigators argued that it modulated normalized damage. Alternatively, papers published during the hyperactive 2005 hurricane season (Emanuel 2005, Webster et al. 2005) fostered the perception that the monumental destruction and mortality during the 2004 and 2005 seasons were immediate consequences of anthropogenic global warming. These interpretations, and virtually any others that one might advance, have been extremely controversial (e.g., Mann and Emanuel 2006, Landsea 2007, Pielke 2005, Vecci et al., 2008).

Just as the lack of a secular trend in damage argues against a century-scale increase in the hurricane threat attributable to Global warming, so too does the lack of a low-frequency “red-noise” component in the damage spectrum corresponding to the AMO undermine meaningful short-term (~5 yr) prediction of hurricane hazards.

By contrast, the spectrum of ACE (Accumulated Cyclone Energy, summed squares of the 6-hourly maximum winds in HURDAT) seems to exhibit a 2nd harmonic peak (i.e., 64-year period). Since the 128 year record spans only two cycles, characterization of this spectral component is dubious. The spectrum itself looks like a textbook case of faulty detrending (e.g. Warner 1998). Moreover, a 500 yr proxy record of the AMO (Gray et al. 2004) resembles low-pass filtered white noise, not a monochromatic oscillation. Thus, one would not necessarily expect that a longer sample of this, or of any time series modulated by the AMO, would exhibit a similar the 64-year periodicity.

References:

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

Emanuel, K. A., 2005: Increasing destructiveness of tropical cyclones over the past 30 years. Nature, 436, 686-688.

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

Gray, S.T., L.J. Graumlich, J.L. Betancourt, and G.T. Pederson, 2004: A tree-ring based reconstruction of the Atlantic Multidecadal Oscillation since 1567 A.D. Geophys. Res. Lett., 31:L12205, doi:10.1029/2004GL019932. Landsea, C. W., 2007: Counting Atlantic tropical cyclones back to 1900. EOS, 88(18), 197-208

Mann, M. E. and K. A. Emanuel, 2006: Atlantic hurricane trends linked to climate change, EOS, 86, 233, 238 & 241.

Pielke, R. A., 2005: Are there trends in hurricane destructiveness? Nature, 438, E11.

Pielke, R. A., J. Gratz, C. Landsea, D. Collins, M. Saunders, and R. Musulin, 2008: Normalized hurricane damages in the United States: 1900-2005. Nat. Haz. Rev., 9(1), 29-42.

Vecchi, G. A., Swanson, K. L., and Sodon, B. J., 2008: Whither hurricane activity? Science, 322, 687-689.

Warner, R. M., 1998: Spectral Analysis of Time-Series Data, p. 168, Guilford Press, New York, NY.

Webster, P. J., Holland, G. H., Curry, J. A., and Chang, H. R., 2005: Changes is tropical cyclone number, duration and intensity in a warming environment, Science, 309, 1844-1846.