Bayesian wind fields estimates from C-band SAR without NWP prior

The accurate estimation of kilometer-scale ocean surface wind fields is a cornerstone of modern marine meteorology, impacting weather forecasting, offshore wind energy development, and maritime safety. Synthetic Aperture Radar (SAR) remains a primary tool for these estimations due to its high spatial resolution and sensitivity to sea surface roughness. Traditionally, wind retrieval from SAR data relies on Geophysical Model Functions (GMFs) that relate the Normalized Radar Cross Section (NRCS) to wind speed and direction. However, because SAR sensors typically operate with a single fixed antenna, the inversion process is inherently under-determined, frequently requiring ancillary data from Numerical Weather Prediction (NWP) models to resolve wind direction ambiguities. These priors can introduce errors due to spatiotemporal lags or the underestimation of extreme events. To address this, more SAR observables can be exploited to gain independence with respect to model a priori. Here, we focus on the Co-Cross-Polarization Coherence (CCPC), defined as the complex cross-correlation between co-polarized and cross-polarized channels, has emerged as a valuable observable to supplement traditional NRCS measurements.

Recently, the CCPC has been formulated as piecewise log-normal function optimized on an extensive dataset of over 25k Sentinel-1 Interferometric Wide Swath (IW) observations. This modeling ensures the model remains well-behaved and physically consistent even for wind regimes higher than 20 m/s. The utility of the new PGMF-2 has been demonstrated through a Bayesian inversion scheme that integrates the co-polarization, the cross-polarization, and the real and imaginary parts of the CCPC. Monte Carlo simulations confirm that the unique odd-symmetry of the CCPC provides critical directional constraints that complement the even-symmetry of the NRCS. This combination enables wind direction retrieval without the need for NWP priors, particularly in the 7 to 15 m/s range, where the SAR-only inversion outperforms traditional NWP-based methods. 

Moving on from simulation, we applied the inversion on real-world observations. Though a prior remain necessary for some wind field configurations, notably when the wind direction is parallel or ortogonal to the satellite track, as the CCPC is zeroed under these directions, the SAR-retrieved field is coherent with both NWP and scaterrometer data. In addition, the method proposed a unified methodology to introduce new parameters, such as the Imax and the streak direction, to further constrain the SAR-inverted wind field. Through the residual of each component, it also provide a quantitative evaluation of the inversion quality, which is primordial for the detection of pathoological cases and provide warning for the users where the SAR data is noised

This research highlights the potential of using additional SAR observables
CCPC to enable autonomous, high-resolution SAR wind field mapping, which is essential for monitoring rapidly evolving extreme weather systems and optimizing offshore energy resources. This opens new wind retrieval perspectives for current and future SAR missions such as Harmony, ROSE-L and Sentinel-1NG.