A New Approach to Visualizing Satellite Microwave Imagery of Tropical Cyclones

Sequential satellite microwave images of tropical cyclones readily lend themselves to morphing applications, which allow for a straightforward visualization of the storms' structural trends. P olar and other low-Earth-orbiting (LEO) satellites provide critical data for use in weather analysis and forecasting. However, the temporal sampling issue leads to the common problem of interpreting sequential images spaced at irregular and sometimes very lengthy time intervals. LEO microwave imagery of tropical cyclones is one of the most extreme cases of this problem, because not only are the weather systems moving, but also the LEO spatial coverage gaps are largest in the Tropics. As a result, the temporal image spacing can range from less than 30 min to more than 25 h, although the more typical spacing tends to vary between 3 and 6 h. Unlike geostationary satellite imagery that is updated frequently, forecasters often acknowledge the difficulty of mentally piecing together a recent history of a tropical cyclone from these irregular LEO datasets. Specifically, the time sensitivity of the interpretation process adds uncertainty to their conclusions on, for example, short-term structural changes. In tropical cyclone applications, the LEO microwave imagery can provide unique information on convective organization, secondary eyewall formation, and eyewall replacement cycles, all of which is critical information when trying to forecast intensity changes (Cecil and Zipser 1999; Jones et al. 2006; Houze et al. 2006). This information is usually lacking in the geostationary channels because of obstructing cloud tops. Clearly, the analysis of tropical cyclones undergoing rapid fluctuations in structure and intensity should benefit greatly from a time-continuous visualization of the LEO microwave imagery. Our approach is to blend sequential individual LEO microwave images together through a " morphing " process, which is the digital manipulation of two or more images to make them appear to change into one another naturally. This approach has been explored in the field of digital image processing for well over a decade [see Ruprecht and Müller (1995) and Wolberg (1998) for good overviews on morphing]. It has been