Tropical Cyclone Classification Using Crowd Sourcing

Hurricane Wilma
We are adopting a tropical cyclone intensity classification scheme for use with thousands of people


Tropical cyclones (TCs) are poorly observed since they develop and spend most of their lifetimes over open ocean and away from dense observing networks. In order to provide an estimate of TC intensity in the absence of observation, an objective algorithm called the Dvorak technique was developed in the 1970s. The technique returned an estimate of maximum TC wind speeds based on the analysis of a single Infrared (IR) satellite picture. Although the Dvorak technique has been improved and adopted by all of the global TC forecast centers, it is well known that there can be large differences interagency differences in storm intensities (e.g. see IBTraCS). Some Dvorak-related reasons for these differences include the subjective application of certain Dvorak rules, different interpretations of the cloud patterns, and/or the application of modified Dvorak rules for particular basins.

The end result is a heterogeneous tropical cyclone intensity record that cannot be used with confidence for trend detection. The goal of this project is to provide new estimates and uncertainties of intensity in the recent TC record through the use of crowd sourcing. Loosely defined, crowd sourcing is the participation of the general public in research projects that are too large in scope to be completed by specialized scientists in a reasonable amount of time. In order to reanalyze TC intensities from 1978-2009, it would take a team of 5 people working 8 hours a day about 23 years. If 5,000 people work on the problem, it can be done in 1-2 years.

In collaboration with the Citizen Science Alliance, we have transformed a version of the Dvorak technique into a web interface that presents users with a simple set of questions. The user responses to these questions (e.g. "What color surrounds the eye?") will allow us to determine an intensity estimate of the TC at that particular time. The application of a modified version of the Dvorak rules post-experiment will yield a homogeneous TC intensity dataset complete with information about each storm including the uncertainty in the intensity (based on the variance of the public classifications). Give it a try at Cyclone Center.

Related Links (Project Participants):

Related Journal Articles:

Related Conference Papers/Web Articles:

  • Hennon, C.C., 2014: Crowd Sourcing Science. International Workshop on Tropical Cyclones-VIII, World Meteorological Organization, Special Focus Session 1c, Jeju, South Korea.

  • Hennon, C.C., 2013: Improving the Tropical Cyclone Climate Record", available at
Related Posters:

Related Presentations:

North Carolina's Public Liberal Arts University