### Abstract

- Top of page
- Abstract
- 1. Introduction
- 2. Data
- 3. Methodology
- 4. Results
- 5. Discussion and Conclusion
- Acknowledgments
- References
- Supporting Information

[1] QuikSCAT data of near-surface wind vectors for the years 1999–2008 are used to create a climatology of tropical cyclone (TC) size, defined as the radius of vanishing winds. The azimuthally-averaged radius of 12 ms^{−1} wind (*r*_{12}) is calculated for a subset of TCs (*N* = 2154) whose centers of circulation were clearly identifiable via subjective analysis of the QuikSCAT-analyzed wind field. The outer radius, *r*_{0}, is determined from *r*_{12} using an outer wind structure model that assumes no deep convection beyond *r*_{12}. The global median values of *r*_{12} and *r*_{0} are 197 km and 423 km, respectively, with statistically significant variation across ocean basins. The global distribution of *r*_{12} is found to be approximately log-normal, the distribution of *r*_{0} is quantitatively much closer to log-normal, and the improvement in fit between *r*_{12} and *r*_{0} is attributed to the combined effect of the nature of the model employed and the paired distributions of *r*_{12} and *f*. Moreover, the normalization employed by Dean et al. (2009) is found to weaken rather than improve the log-normal fit. Finally, within a given storm, both *r*_{12} and *r*_{0} tend to expand very slowly with time early in the storm lifecycle and then becomes quasi-constant, though significant variance exists across storms.