Wednesday, 26 January 2011: 10:45 AM
607 (Washington State Convention Center)
An organized and objective method for testing any centroid-based storm cell tracking algorithm is investigated. Storm cell tracking is an important component for storm evolution studies, real-time nowcasting and severe weather warning systems. Several algorithms are currently used for these purposes, however, efficacy studies for these tracking algorithms have been of marginal use. Limitations such as the number of case studies, arbitrary metrics, and subjective analyzes restrict the scope and applicability of the methods used to determine efficacy. An objective, generalized method for testing and scoring any tracking algorithm is necessary for quantitative comparisons. It would also have an added benefit of allowing experiments to be constructed to study the behavioral characteristics of the algorithms in different environmental scenarios.
A system for testing and scoring any centroid-based storm cell tracking algorithm is described. The system simulates the track of the centroid of many storm cells in configurable scenarios. Tracking algorithms are then applied to the simulated data and the system uses this truth data to score the results. Several track scenarios, both idealized and realistic, are examined. The utility of this testing system for the purpose of determining an optimal tracking algorithm is discussed.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner