Résumé
Tropical cyclone (TC) internal dynamics have emerged over recent decades as a key to understand their intensity variations, but they are difficult to observe because they are sporadic, multiscale, and occur in areas of very strong wind gradients. The present work aims at describing the internal structure of TCs, as observed with newly available satellite synthetic aperture radar (SAR) wind products, and at evaluating relations between this structure and the TC life cycle. It is based on a unique dataset of 188 SAR high-resolution (1 km) images, containing 15-47 images by intensity category. An extraction method is designed to retrieve and characterize the TC radial profile, its azimuthal degree of asymmetry, and the energy distribution in the eyewall and maximum wind areas. Vortex contraction and sharpening of the eyewall wind radial gradient with increasing TC intensity are observed, as well as a symmetrization of energy distribution around the vortex. Eyewall high-wavenumber structures show a dependence on the life-cycle phase, supporting previous findings discussing the vortex rapid evolution with onset and propagation of eyewall mesovortices and associated vortex Rossby wave generation. A machine-learning approach highlights that the eye shape and eyewall radial wind gradient fine-scale dynamics have the potential to improve the statistical prediction of TC intensity variations relative to the sole use of vortex-averaged parameters and synoptic information. The high-resolution radial and azimuthal coverage provided by SARs make these acquisitions a very valuable tool for TC research and operational application.