Contents I. Introduction
Sea surface wind speed is an indispensable parameter for the studies of the other parameters [1]–[3], i.e., wave, current, and the coupling of oceanic and atmospheric systems. Likewise, it is a crucial factor for the studies of oil spill propagation [4] and vessel detection [5]. In particular, surface winds play an important role for modeling radar backscattering from sea surface. Among surface wind sources, i.e., in situ measurements from meteorological buoys/stations, numerical weather prediction models, scatterometers, synthetic aperture radars (SAR) at L-, C-, and X-band, are more and more exploited, since they can offer wind speed estimates at a high spatial resolution and in most meteorological conditions (day/night and cloud). In general, SAR-derived wind speed estimation has been principally based on empirical (EP) models, e.g., CMOD functions [6]–[9], since such approach is quite simple, and thus effective in obtaining wind speed estimates rapidly. Few studies mention wind speed retrieval through physical models, i.e., Bragg-scattering resonance [10], [11] and approximation models [12]–[14], since those models associated with many oceanic parameters (wave spectrum and sea water permittivity) are complex, and thus are not relevant for wind speed estimation from SAR data. Nevertheless, through theoretical calculations, one can understand more deeply physical processes between radar backscattering and sea surface roughness which is strongly influenced by near-surface winds. In spite of two different approaches, the comparison between EP and physical models is necessary and relevant for understanding which elements have not been taken into account in radar backscattering modeling, and then it permits to improve the calculations of physical processes (i.e., reflection and dispersion) between radar signal and sea surface. The latter is particularly significant for the design of radar sensors [15]. For instance, the design of the cyclone global navigation satellite system [16] is based on the hybrid model which is constructed from the physical and EP calculations of radar backscattering from the sea surface. Surface wind speed is the selected parameter for the comparison of EP and physical models due to its important role in modeling radar backscattering from sea surface roughness. In addition, wind speed estimates can be easily compared to available in situ measurements. Though the comparison of wind speed estimates based on EP and physical models has been reported in [17]–[19], it is not still adequate to conclude the performance of the physical models in radar backscattering modeling, since the approximation models [12]–[14], especially resonant curvature approximation (RCA) [14], have not been mentioned. Hence, for an overview, this paper proposes the comparison of wind speed estimate-based EP and physical approaches including the approximation models.
