High-Resolution 2-D Cloud-Top Advection From Geostationary Satellite Imagery

This paper introduces a hierarchical mesh-based tracking algorithm designed to retrieve two-dimensional cloud-top advection to near-pixel resolutions from geostationary satellite imagery. The algorithm associates each cloudy pixel in the current image with pixel locations in the previous and subsequent images corresponding to the equivalent location within the same cloud. Such a product resembles a high-resolution cloud-derived wind field but explicitly incorporates the effects of cloud growth, deformation, and dispersal. Within the algorithm, cloud advection is represented by the topological transformation of a rectangular mesh. An iterative procedure progressively refines the spatial resolution of an initially coarse mesh until image pixel resolution is reached. The algorithm has been validated using Geostationary Operational Environmental Satellites (GOES) infrared imagery