Tuesday, September 18, 2007

Trenberth on hurricanes making landfall -- clarity emerging?

Climate Feedback: The situation on how hurricanes have changed and will likely change in the future are outlined in my recent Scientific American article, but may seem as murky as ever to the public although clarity is actually emerging.

A recent news report in the Press Register outlines some sources of confusion related to just how well the past record is known. It cites work by Landsea that relied on numbers of land-falling storms as a way to calibrate the Atlantic hurricane record, and which concluded that there may be an undercount of 3.2 storms per year prior to 1966. The most recent Eos Transactions has two articles that follow up and point out why use of land-falling storms is misleading [Holland 2007; Eos 88 (36) 348-349] and that the conclusions of increased activity do not change anyway [Mann et al., 2007; Eos 88 (36) 349-350].

The Holland article points out that there are good reasons why the fraction of storms making landfall should change, both because of natural variations and especially if the climate changes. The Mann et al. article adopts the Landsea-suggested changes for the past as a “what if” test and goes on to show that even a substantial underestimate of early 20th century storms does not change the significance of the increase in activity since 1994. Nor does it change the strong relationship with observed sea surface temperatures (SSTs) in the region; the SSTs have a much more reliable observational record and have clearly increased.

Surprisingly, none of these studies refers to what seems to me to be the most definitive analysis of the likely missed storms in the historical counts by Chang and Guo (2007) in which they analyze in detail the actual ship tracks in the past compared with modern tropical storm tracks. To quote their main conclusions: “It is estimated that the number of tropical cyclones not making landfall over any continent or the Caribbean may have been underestimated by up to 2.1 per year during 1904–1913, with this number decreasing to 1.0 per year or less during the 1920s and later decades. Our results suggest that the characteristics of North Atlantic tropical cyclone track statistics might have changed during the 20th century.”

In 2007 the tropical storm season has been fairly normal in many respects up to now. Only 3 hurricanes have been recorded (versus average 3 to 4) but two were category 5 storms, and that is highly unusual. Forecasts of hurricane activity by NOAA and Bill Gray continue to forecast substantially above normal activity in the Atlantic. To me, observing the events thus far, the incredibly intense convective activity in the Indian Ocean from May to July was an important and totally overlooked factor. The subsequent heavy rains and flooding in India and China were no doubt related. The fact that Atlantic hurricane activity is influenced by events in the Indian Ocean seems to be overlooked by the hurricane forecasters.... Kevin Trenberth

No comments: