Ocean Remote Sensing Using GNSS-R - GNSS Remote Sensing: Theory, Methods and Applications - page 232

Fig. 9.9 Example of a particular numerical realization of the surface topography. It has been
made with the wind-driven waves' spectrum in Elfouhaily et al. ( 1997 ), fully developed, 6 m/s
along the X-direction. A swell of p = 400 m and SWH 1 and 50 ı direction has been added

using Pierson-Moskowitz spectrum. A close-up is displayed on the right panel .X-andY-axis
annotations are given in meter

D

where a random arbitrary phase is added to each wavenumber, . k /, and the real part
R is taken. It is worth checking the normalization (see Eq. 9.17 )aswellastheMSS

to detect any possible problem in the definition of the used wavenumber spectra.
Figure 9.9 presents a numerical realization of a surface topography with both
wind-driven waves and swell obtained from Eq. 9.30 ,using( Elfouhaily et al. 1997 )
spectrum for wind-driven waves, and Pierson-Moskowitz for long peak wavelength
to model the swell.
Because of the speckle, which induces noise proportional to the power (Sect. 8.7 ),
the scattering off one single realization is not significant, but large ensembles are
required.

9.2.2 Retrieval Approaches

This section compiles some of the GNSS-R Ocean surface roughness inversion
techniques applied so far, mostly to extract surface winds and MSS. In particular,
some approaches extract the wind speed, some others wind vector; isotropic MSS;
2D-MSS; and discrete and non-parameterized slopes' PDF. Other studies do not
seek the inversion into an Oceanographic parameter, but a direct link to another
remote sensing technique: roughness corrections required in L-band radiometry in
order to properly extract surface salinity.
The techniques are briefly described below, and terminology for GNSS-R
observables follows the one given in Sect. 8.10 :

DM-fit: After re-normalizing and re-aligning the delay-waveform, the best fit
against a theoretical model gives the best estimate for the geophysical and