Powell & Black studies of dropsonde and bouy data suggest that 63% to 73% to be much more reasonable ratios that what is popularly promulgated. NHC has often come under criticism with their wind speed estimates. However, in the case of hurricane Mitch (1998), the wind speed profile was unique. More recent studies done by Hock & Franklin using GPS based dropwindsonde in 1999, gave the first clear cut view of eyewall wind profile (with accuracy of 15' and 1-4MPH).
If I'm reading that NHC document correctly, the .73 pertains to max winds (found at 1600 feet), with the .91 being applied to flight level winds (10,000 feet), which apparently are not max, but have been found to be typically 20 per cent lower than those found at 1600 feet. Which would explain why the NHC uses a .91 multiplier from flight level.
My experience in hurricanes is that the surface anemometers are usually destroyed at some point before greatest wind speeds can be recorded.