The areal distribution of the deaths based on Appendix 1 is
shown in the following table; but, as described below, these totals
indicate losses that are likely significantly lower than the actual losses.
Areal distribution of deaths due to Atlantic tropical cyclones listed in
Appendix 1. Totals are rounded.
|Greater Antilles||45,000 (29%)|
|Offshore Losses||35,000 (22%)|
|Lesser Antilles||35,000 (21%)|
|United States mainland (Galveston storm: 8,000)||25,000 (16%)|
|Mexico and Central America||20,000 (12%)|
|Elsewhere (Azores, Bahamas, Bermuda, Canada, Cape Verde Islands, South America, Ireland)
The disposition of the many casualties from shipwrecks near shore
into offshore versus land losses is not certain.
It is interesting that over 90% of the offshore losses
occurred more than 200 years ago (before 1790), as did all 12
offshore losses of more than 1,000 people. For the continental
United States, the Galveston storm was responsible for about one-third of the deaths
(using data in Appendix 1 only).
The areal totals indicate a large death toll across the region.
They do not, however, adequately reflect the threat of the individual intense
hurricane. We note that the five
tropical cyclones at the top of
Appendix 1 (1780 Great Hurricane, 1900 in
Galveston, 1974 Fifi, 1930 in Dominican Republic, 1963 Flora)
account for about one-third of all the deaths over the past 500
years in storms for which quantitative data on deaths has been
found. In fact, the 10 deadliest storms, while representing less
than 5% of the cases in Appendix 1 and less than 0.2% of all
tropical cyclones since 1492, account for almost one-half of the
deaths indicated in Appendix 1.
These statistics point to the tremendous repercussions that
small track changes have had (and will have) on population centers
at risk from a potentially pastdeadly storm. A shift of about 50 miles
in the track of the 1900 Galveston hurricane could have meant far
fewer deaths on that vulnerable island and (hence) overall. (This
distance is comparable to the current average NHC 24-hour "across"
track forecast error.) On the other hand, because of the growing
population, there is an increasing number of highly susceptible
regions which, only so far, have escaped such a catastrophic event
(e.g., Sheets 1990). Damage statistics also illustrate this point.
In 1992, Hurricane Andrew caused around $25 billion damage in South
Florida (Rappaport 1993). An estimate indicates that a 20 mile
northward shift of Andrew's track would have resulted in two to
three times that much damage (Doig, 1992). Alternately, a
southward shift of about 40 miles could have resulted in a
negligible monetary loss to mainland Florida (but additional
problems, including possible loss of life, for the less-populated
The total number of deaths associated with Atlantic tropical
cyclones of the past five centuries is likely much larger than
implied by the data in Appendix 1. While it is a statistic that
cannot be specified with confidence, a range for the total loss can
be estimated. Appendix 1 provides a starting point and an
underestimate of the total loss. Using the first number in the
column on deaths for each case (except using 8,000 for the
Galveston storm), the total number of deaths obtained from the
table is around 160,000.
To this, we add several considerations.
(a). Many of the entries in Appendix 1 are minimum estimates
(note the numerous or > symbols).
(b). We chose the first (largest) total in each case for
Appendix 1 (except for the Galveston storm). In some cases, a
smaller total could be more accurate.
(c). Some storms with footnote c in Appendix 1 may not have
been tropical cyclones.
Overall, consideration (a) probably dominates. We estimate
that the total for Appendix 1 is around 200,000.
The number of entries in Appendix 2 is smaller than in
Appendix 1 and some of these cases probably did not result in 25
tropical cyclone deaths. These storms are probably responsible for
an additional number of deaths that is considerably less than 200,000.
(a). Greater than 25 deaths. We believe that most disasters
responsible for very large losses are already documented in the
Appendices, and that the remaining cases probably contribute less
(b) Less than 25 deaths. Based on information in Monthly
Weather Review, the number of deaths associated with this item in
the past 50 years is about 575. If this data is representative of
the entire study period, then these losses are less than 10,000.
Based on the above, we speculate that the number of deaths in
Atlantic tropical cyclones from 1492-1994 is between one-third and
one-half million. Factors contributing to the uncertainties noted
above include relatively few references to losses in Mexico and
Central America and incomplete information about losses from
Spanish ships in the 1500's-1700's and to slaves and natives of the
region. There are sources that could provide more definitive
information, including old newspapers reviewed in a more systematic
manner. This phase of the research is underway.