Shen-Shyang Ho, Ashit Talukder
To track a cyclone using a single orbiting satellite in a continuous manner is impractical as it has limited spatial and temporal coverage. One solution is to use multiple orbiting satellites for cyclone tracking. However, data from some orbiting satellites do not provide features as useful as other satellites in identifying cyclones. Moreover, satellite data containing strong cyclone discriminating features is affected by coarse temporal resolution and object occlusion while satellite data containing weak cyclone features does not have positive examples for cyclone identification. In this paper, we propose a methodology for spatial-temporal knowledge transfer to enable cyclone identification and detection using data with weak features in a multiple data sources setting. This approach also minimizes the negative effect of coarse temporal resolution and occlusion when only the satellite data containing strong cyclone discriminating features is used. Experimental results are presented to demonstrate the feasibility and usefulness of our knowledge transfer approach for cyclone tracking.
Subjects: 12. Machine Learning and Discovery; 1. Applications
Submitted: May 5, 2008