Sea surface wind speed retrieval through Sentinel-1 wavelength cut-off
- Paper number
IAC-16,B1,IP,17,x33951
- Coauthor
Dr. Giuseppe Grieco, Matera Space Geodesy Center, Agenzia Spaziale Italiana (ASI), Italy
- Coauthor
Prof. Maurizio Migliaccio, Università degli studi di Napoli Parthenope, Italy
- Coauthor
Dr. Marcos Portabella, Institut de Ciències del Mar, Spain
- Year
2016
- Abstract
The retrieval of the sea surface wind speed (U) by Synthetic Aperture Radar (SAR) images is still an open question for some aspects. The retrieval of $U$ through the wavelength azimuth cut-off ($\lambda_{C}$) is a problem that has been addressed in literature. Some studies evidence a significant empirical dependency of $\lambda_{C}$ on the significant wave height ($H_{s}$) and on $U$ but the feasibility of the retrieval of $U$ through $\lambda_{C}$ has not yet been demonstrated, even if suggested and attempted. In this study, we evidence that the correlation between $\lambda_{C}$ and $U$ is only high for fully developed sea states. Therefore, we focus our attention on the retrieval of $U$ through $\lambda_{C}$ only in such situations. The fitting procedure of a Geophysical Model Function (GMF) that relates $\lambda_{C}$ to $U$ in fully developed sea states is presented and discussed. The fitting procedure is carried on by taking benefit of a dataset of multi-look images acquired with varying incidence angles in the range between 20° and 45° by the new C-band SAR Sentinel-1 of the European Space Agency (ESA). These images have been co-located with U and Hs from the European Centre for Medium Range Weather Forecast (ECMWF) operational model output. The results of a retrieval exercise are presented and discussed. The retrievals of $U$ are validated against the $U$ measurements of the Chinese scatterometer of the satellite mission HY-2A (HSCAT). The agreement with the scatterometric measurements is within 1 m/s for wind speeds higher than 15 m/s. $\lambda_{C}$ strictly depends on the range to SAR platform velocity ratio and its computation does not need a calibrated Normalized Radar Cross Section (NRCS). Therefore, $\lambda_{C}$ computed from different SAR systems should be similar in principle, given that the correction for the acquisition geometry is applied and that the differences due to the SAR wavelength are negligible. In this study we compare $\lambda_{C}$ computed from COSMO-SkyMed SAR images with from $\lambda_{C}$ Sentinel-1 images in similar $U$ and $H_{s}$ conditions and the differences are discussed.- Abstract document
- Manuscript document
(absent)