Is the number of North Atlantic tropical cyclones significantly underestimated prior to the availability of satellite observations?

The number of North Atlantic tropical cyclones that may nave been undetected before satellite observations are available is estimated by passing the cyclone tracks taken from 1976 to 2005 through ship observations from 1900 to 1965. The probability of detection is equated to the probability that the ships would have made wind observations of 18 m/s or higher had the tropical cyclones been present during the earlier years, based on the probability computed from actual wind observations around tropical cyclones during the satellite era. It is estimated that the number of tropical cyclones not making landfall over any continent or the Caribbeans 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.