Cyclone Landfalls, USA, 1851-2018

The list of cyclone landfalls in the USA, 1851-2018 can be found at the AOML_HRD Site (NOAA, Hurricane Research Division). From such a list I've extracted both usa.txt and usa3.txt (the first one concerns all the events, while the other concerns categories 3,4,5).
Another list exists, from Blake e Gibney,2011 where data are different, in spite of the same derivation; in practice the last list uses a different classification of the cyclones.

I don't know what the correct list may be, so I'll plot both lists but pls note that the AOML_HRD is updated on July 31, 2018, while Blake & Gibney paper is a 2011 NOAA Technical Memorandum (NWS NHC-6).

  • I regret so much to highlight that the following plot, published by Roy Spencer at his own site, is wrong and misleading because
    1)

    The negative slope in figure 1 is an arctifact depending on the cherry picking and the use of incomplete data.
    Such a trial is ingenous and not useful because, as plots 2) and 3) show, the correct slope is pratically zero, giving a rise of 0.96 and 1.1 events in 160 years, with an almost 100% relative error, for plot 2) and 3) respectively. We don't need a bad use of the available data; we can state, in perfect good faith, that the major hurricane landfall series doesn't show any growth in the last century and half.

    A reader of my post on ClimateMonitor informed us that Roy Spencer previously published the cyclone rainfalls within the whole (17) decades here. He also justify the choice to plot 9th through 17th decade (here presented along with the complete plot) as a comparison, in such a way:
    Why did I pick the 1930s as the starting point?
    Because yesterday I presented U.S. Government data on the 36 most costly hurricanes in U.S. history, which have all occurred since the 1930s. Since the 1930s, hurricane damages have increased dramatically. But, as Roger Pielke, Jr. has documented, that's due to a huge increase in vulnerable infrastructure in a more populous and more prosperous nation.

    This may be a correct explanation for the cherry-picking used here in conjunction with the complete plot. Nevertheless, it cannot explain the lack of information associated to the last-published plot. It appears as an uncorrect and misleading choice of particular decades, in order to confirm his own ideas. Two rows of a newly repeated explanation would have been appreciated, while the actual choice appears to be a serious fault in Spencer credibility (and I'm very,very sorry to write that, really!).

    Also, the use of the 17th decade whose numer of cyclones is totally unknown, is justified by:
    ... if we assume the long-term average of 6 storms per decade continues for the remaining 2.5 hurricane seasons, the downward trend since the 1930s will still be a 50% reduction.
    but it appears unsupported and on the same run as the AWG supporters. Not necessary and misleading.


    2)


    3)

    Someone correctly assumes that the use of USA landfall cyclones means a strong limit to the number of events. I agree with them and have used the data of all the cyclones, 1967 through 2017, selected in categories, available in Wikipedia and, given the starting date, relative to the satellite era. The plot of the ≥3 events follows. It can be noted the real positive

    4)

    slope which could bring to the idea of a growing number of extreme events in the last 50 years, but we note a break point near the years 1994-96 in the graph. So, two linear fits: before and after 1995, highlight two very different regimes of the series: before 1995 the cyclones number was constant on average; after 1995 the climate shift (in many European climatic series a climate shift in 1987 is also visible) the (relatively) large number of events tends to fall at a (0.05±0.06) events/yr rate. So, the growing slope derived from the overall fit cannot be considered as a real one, given the two different behaviour of the data.


    Page written: October 18, 2018;     Last updated: October 21, 2018