Measuring Tropical Cyclone Internal Properties from SAR Imagery : Current Capabilities and Perspectives for the Improvement of Intensity Forecasts.

Throughout their life cycle, Tropical Cyclones (TC) undergo strong intensity fluctuations that remain a challenge for TC forecasters. Occurring through an interplay between structural perturbations (as eyewall breaking and distortion or vortex tilting) and restoration processes (as vortex Rossby waves and eyewall contraction or replacement), these fluctuations are inherently multiscale, transient, and asymmetric. Despite their significance, these processes pose challenges in observation and are not routinely used as indicators by meteorological centers. In recent years, Synthetic Aperture Radar (SAR) observations have emerged as novel sources and innovative measurement methods for TC properties. It is indeed the only satellite remote sensing technique that can probe the sea surface and provide estimates of 2D wind fields with kilometer resolution.

Based on a substantial dataset of 188 high-quality TC-centered SAR acquisitions from Sentinel-1 and Radarsat-2 processed at IFREMER since 2015, the present study aims to comprehensively measure internal structure properties derived from SAR-estimated TC surface wind fields and establish connections between these properties and the intensity and life cycle of TCs. An extraction method is designed to retrieve and characterize the radial wind profile and the degree of asymmetry of characteristic azimuthal distributions (eyewall size, maximum wind distribution, etc.). The extracted parameters are shown to exhibit correlations with the intensification regime of TCs, such as U-shaped eyewall wind profiles being significantly associated with more intense and intensifying TCs. A machine learning classification method is applied to assess the impact of internal parameters on the dissociation of intensifying and decaying TCs. High wavenumbers associated with azimuthal distributions of eye shape radius and eyewall radial wind gradient prove to be the most influential contributors, yielding a classification performance 4.5% better than models excluding SAR-extracted parameters. In addition, the refined estimates of the TC wind field structure allows the formulation of a symmetric parametric surface wind field model based on the statistical regression of SAR-measured radial wind profile exponents as expressed in various existing parametric models. An enhanced SAR-fitted TC parametric model is proposed for improved risk assessment.

As SAR only provides instantaneous diagnoses of TC internal structure with almost no temporal continuity, a complementary modeling dataset is built to evaluate the potential of more regular acquisitions for TC intensity predictability. Through realistic dynamical simulations, the temporal variations of SAR- equivalent parameters are examined and cross-compared with intensity variations. A novel metric is introduced to measure the transition of azimuthal distributions between phases of concentrated and distributed asymmetry, by computing the joint evolution of low and high wave numbers extracted from the azimuthal spectrum. This metric shows connections to short-scale intensity variations. On average over several events simulations, the fueling of high wavenumber asymmetries distributed around the ring of maximum winds is shown to precede phases of rapid (re-) intensification by 5-6h, while the concentration of asymmetry in wavenumbers 1 and 2 leads to intensity weakening. The aforementioned classification method is reapplied and highlights that the classification of intensification phases (i.e., intensification or weakening) can be improved by at least 11% (thus reaching ∼75% of performance) when accounting for the evolution of the radial wind gradient and of the power transfer between low and high wave numbers in the ring of maximum winds, relative to the sole use of vortex- averaged parameters.

These results advocate for a broader utilization of SAR observations in the context of TC study and forecasting. The extensive SAR database provides unique insights into the variability of the surface wind field and holds the potential to improve our anticipation of intensity variations—a significant challenge in modern TC forecasting.