For many tropical cyclones, the Dvorak technique is the only operational tool available to estimate the cyclone’s maximum sustained wind speed. However, when reconnaissance aircraft data are available, there are often large differences between the Dvorak satellite-based estimates and those determined from the reconnaissance data. In 1997, reconnaissance aircraft began releasing GPS dropwindsondes into the eyewall or maximum wind band of tropical cyclones. These observations have improved our ability to estimate the maximum surface winds of a tropical cyclone, both by direct measurement and through more accurate interpretation of aircraft flight-level winds. As a result, it is likely that NHC “Best-Track” wind data since 1997 are more accurate than in prior years.

Using reconnaissance-based best track data, Brown and Franklin (2002) found that the RMS “error” of the Dvorak satellite estimates over the period 1997-2001 was 12.1 kt. It was also determined that 25% of the satellite intensity estimates had apparent errors of 14 kt or greater. The present paper represents a continuation of the Brown and Franklin study, and attempts to quantify some of the likely sources of error in the Dvorak estimates. Some specific issues to be addressed include: are satellite intensity estimates of weakening tropical cyclones accurate? If not, should the Dvorak rules be amended. What affect does the translational speed or the latitude of the tropical cyclone have on satellite estimates?