Determining Uncertainty in Historical Tropical Cyclone Intensity Analyses

The historical global tropical cyclone (TC) intensity record is overwhelmingly influenced by subjective satellite intensity estimates from a small number of analysts. When multiple forecast agencies analyze the same storm, significant differences in the maximum sustained wind speed periodically arise. These discrepancies propagate across time in the best-track records and have produced conflicting decadal trends in TC intensity, particularly in the western North Pacific.

Cyclone Center is a crowd sourcing effort that applies a Dvorak-like intensity algorithm to data collected from citizen scientists examining infrared satellite images of TCs. The project is uniquely positioned to provide uncertainty information since at least ten unique users classify each TC image. Cyclone Center produces 3-hourly TC wind speed along with measures of uncertainty for each image and storm. The uncertainty is based on the homogeneity of user classifications and the chosen cloud patterns.

We will present several examples of storm classifications and intensity uncertainty. We believe that the data will be useful for identifying targets for reanalysis, resolving disparate trend analyses, and increasing confidence in risk-based systems that use historical intensity data such as catastrophe models.