Tuesday, 8 January 2013: 2:30 PM
Room 12B (Austin Convention Center)
This talk will discuss the SphereBlur package, written in Python and available in the Ultra-scale Visualization Climate Data AnalysisTools (UV-CDAT) environment. SphereBlur provides a flexible multi-scale analysis toolkit for climate data based on linear scale space. Scale space methods, common in image processing, draw upon the well-studied physics of diffusion to obtain a multi-scale representation of data. A simple extension of these methods to the sphere provides flexible analysis tools for climate data. We use this framework to evaluate model performance at multiple spatial scales and to design filters to isolate scales of interest. We show how these methods can be used to simultaneously detect points and scales of interest in data, and to track the appearance and evolution of features such as corners, edges, and blobs in observational and model data.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner