Automated implementations of the Advanced Objective Dvorak Technique (AODT) developed at CIMSS require a robust estimate of the tropical cyclone's center of rotation. Contemporary automated estimates of the center of rotation using IR imagery are often hampered by high clouds covering the eye, or by misleading, high brightness-temperature "moats" (false eyes) in close proximity to the eye. This presentation describes new approaches to resolve the center of rotation of a tropical cyclone, with algorithms that can be applied to either IR or microwave imagery. The first algorithm finds a "spiral center" indicated by an organized spiral structure in the image gradient vector field. This approach is strongly unaffected by the presence of "moats". The second algorithm uses a variation of the Hough transform to resolve a circle-type pattern in either a clear or partially obscured scene, and can be used to discriminate an eye from a "moat". When used together, these algorithms lead to a substantially more robust center-fixing operation in the CIMSS Automated Optical Dvorak product that is less dependent on persistence or short-term forecasts of the tropical storm center.

To assist in the center-fixing methods, the problem of time gaps in the microwave imagery is ameliorated by image "morphing" operations, which attempt to create a seamless, realistic transition between images that uses matching features in the separate microwave images to create feature-preserving transformations in high temporal resolution. Applications of these morphing techniques to successive SSMI passes over tropical cyclones will be discussed.