IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (Jan 2024)
Evaluation of SAR-Based Sea State Parameters and Roughness Length Derivation Over the Coastal Seas of the USA
Abstract
This article presents comprehensive validation of specific sea state parameters (SSPs) and synthetic aperture radar (SAR)-derived wind speeds (${{u}_{\text{SAR}}}$). The article introduces a novel approach to retrieving roughness length (${{z}_0}$) based on wave steepness, following the retrieval of the short wavelengths necessary to estimate ${{z}_0}$. The SAR onboard the Sentinel-1 (S1) satellite that was used specifically in the interferometric wide swath mode (IW) data. The data were processed using the extended version of CWAVE (CWAVE_EX) algorithm for SSPs and CMOD5 for ${{u}_{\text{SAR}}}$. CWAVE_EX was developed especially for coastal waters; the processing chain includes steps for SAR image denoising and eliminating image artifacts. SAR S-1 data inherently exhibit a substantial azimuthal cutoff length due to the data's high satellite altitude and SAR IW resolution. That complicates the retrieving of short wavelengths prevalent in coastal zones and needed to retrieve ${{z}_0}$. The article focuses on the coastal seas of the USA, benefiting from the presence of an extensive network of ocean buoys for validation purposes. The complete SAR S1 A/B archive from 2014 to 2022 was first processed to retrieve SSPs and ${{u}_{\text{SAR}}}$. The validation for significant wave height (${{H}_s}$), second moment wave period (${{T}_{m2}}$), and ${{u}_{\text{SAR}}}$ was performed using in-situ measurements with about 6000 collocations. ${{H}_s}$ and ${{T}_{m2}}$ were compared against the corresponding parameters from hindcast spectral numerical model data with about 380 000 collocations. The comparisons between the retrieved ${{H}_s}$ and ${{T}_{m2}}$ against the in-situ observations and hindcast wave model data yielded a root mean square error (RMSE) of 0.46–0.50 m and 0.9–1.1 s. The RMSE of ${{u}_{\text{SAR}}}$ against in-situ observation was about 2 m/s with a bias of 0.78 m/s. The estimated ${{z}_0}$ values from satellite-driven wave parameters were highly correlated with the ${{z}_0}$ estimated from the in-situ observations, with an RMSE of 0.04 × 10−3 m and a bias of −0.01 × 10−3 m. The article highlights the possibility of using SAR remote sensing data for global mapping of ${{z}_0}$, including coastal effects of local variability in sea state and wind field gustiness.
Keywords
