Contour Detections

In discussions with National Weather Service forecasters and others end users, it became apparent that for improved data presentation quality, scanning to the very leading edges of storms is preferred over focusing specifically on the areas of highest intensity.  Since both of the existing reflectivity-based algorithms used in the closed-loop IP1 test bed are forms of local-maxima finders and are not intended to fully describe the irregular geometries of weather, we decided to introduce data contouring into the closed loop. Contouring provides good geographical description of the location of relevant weather to the MC&C, and the closed-form nature of a contour facilitates the difficult problem of objectifying data out of fields, or more specifically, identifying a 'storm' and maintaining that association over time as the structure evolves. The contour points represent a layer in between the raw field level and the pure object level; less voluminous and more user-centric than moment data, but still retaining enough spatial information to provide accurate description, which higher level abstractions lack. The contour algorithm was officially brought online in mid-2008 as a collaborative effort between the End User, Analysis and Predicting, and Distributing thrusts.

This new approach of representing meteorological features as contours is a significant step towards representing weather in the form of objects. This approach is a basis for new data compression methods and could be the foundation for a new "object-based" display type.

Contour detections overlaid over merged reflectivity image