13 Use of SAR data to Monitor the Greater Agulhas Current
253
are associated with an overall range directed bulk velocity including the mean motion
of the radar-detected wind scatters at the ocean surface and the desired ocean surface
current.
Using Envisat ASAR observations, Chapron et al. (2005) demonstrated the capability to use the Doppler centroid information embedded in the radar signal to map
surface velocity, including wind-generated waves and current, from SAR images.
The difference between a predicted Doppler shift based on precise knowledge of
the satellite orbit and attitude, and the Doppler centroid frequency estimate represents the geophysical Doppler shift experienced from the moving ocean surface. This
geophysical Doppler shift in turn reflects the line-of-sight velocity of the scatterers,
weighted by their contribution to the backscattered power (Romeiser and Thompson
2000; Romeiser et al. 2010). A detailed decription of the retrieval and subsequent
error correction of the geophysical Doppler shift from the ASAR WSM product is
presented in Hansen et al. (2011a). The accuracy of the geophysical Doppler shift is
found to be about 5 Hz. This corresponds to a horizontal surface velocity of about
0.2 m/s at an incidence angle of 40
◦ . As such, the accuracy is still an issue in single
scenes, although temporal averaging has been shown to capture the mean circulation
in e.g. the Agulhas region (Rouault et al. 2010) and in the Norwegian Sea (Hansen
et al. 2011b).
The variation of the Doppler shift as a function of wind speed and radar configurations has been determined by collocating wind measurements at 10 m height from
the European Centre for Medium-Range Weather Forecast (ECMWF), updated every 6 h with 0.5
◦ spatial resolution, with C-band ASAR Doppler anomalies (Collard
et al. 2008). In so doing a global match-up (collocated) database has been established between wind speed and direction from ECMWF with incidence angle and
Doppler measurements from ASAR leading to 277211 collocations for VV polarizations (Mouche et al. 2012). Areas with strong currents have been excluded in order to
avoid leakage of surface current signals into the wind speed-Doppler shift correlation.
From this database an empirical function, named CDOP, that relates the Doppler shift
at C-band to the wind field and the radar configuration was then determined as shown
in Fig. 13.1 (Mouche et al. 2012). The spread indicated by the vertical bars can be
explained by several factors such as the shift in time between ECMWF outputs and
ASAR acquisition, errors in the wind from the model, other geophysical phenomena
such as rain
2 , errors in the Doppler estimates and the non-geophysical corrections
applied to the Doppler centroid. Using the CDOP relationship the residual Doppler
anomalies associated with the surface current can thus be determined with an r.m.s.
error of 5 Hz (equivalent to 0.2 m/s) at a spatial resolution of 8 km (along-track) by
4 km (across-track) as demonstrated by Johannessen et al. (2008) and Hansen et al.
(2011b).
The ascending Envisat tracks yield a more favorable imaging geometry with
respect to the core flow direction of the greater Agulhas Current as shown in
Fig. 13.2. Accordingly, the maximum mean radial Doppler velocity reaches about
2 Rain impacts the roughness and the local wind speed at a scale not resolved by Numerical Weather
Prediction (NWP) models.
253
are associated with an overall range directed bulk velocity including the mean motion
of the radar-detected wind scatters at the ocean surface and the desired ocean surface
current.
Using Envisat ASAR observations, Chapron et al. (2005) demonstrated the capability to use the Doppler centroid information embedded in the radar signal to map
surface velocity, including wind-generated waves and current, from SAR images.
The difference between a predicted Doppler shift based on precise knowledge of
the satellite orbit and attitude, and the Doppler centroid frequency estimate represents the geophysical Doppler shift experienced from the moving ocean surface. This
geophysical Doppler shift in turn reflects the line-of-sight velocity of the scatterers,
weighted by their contribution to the backscattered power (Romeiser and Thompson
2000; Romeiser et al. 2010). A detailed decription of the retrieval and subsequent
error correction of the geophysical Doppler shift from the ASAR WSM product is
presented in Hansen et al. (2011a). The accuracy of the geophysical Doppler shift is
found to be about 5 Hz. This corresponds to a horizontal surface velocity of about
0.2 m/s at an incidence angle of 40
◦ . As such, the accuracy is still an issue in single
scenes, although temporal averaging has been shown to capture the mean circulation
in e.g. the Agulhas region (Rouault et al. 2010) and in the Norwegian Sea (Hansen
et al. 2011b).
The variation of the Doppler shift as a function of wind speed and radar configurations has been determined by collocating wind measurements at 10 m height from
the European Centre for Medium-Range Weather Forecast (ECMWF), updated every 6 h with 0.5
◦ spatial resolution, with C-band ASAR Doppler anomalies (Collard
et al. 2008). In so doing a global match-up (collocated) database has been established between wind speed and direction from ECMWF with incidence angle and
Doppler measurements from ASAR leading to 277211 collocations for VV polarizations (Mouche et al. 2012). Areas with strong currents have been excluded in order to
avoid leakage of surface current signals into the wind speed-Doppler shift correlation.
From this database an empirical function, named CDOP, that relates the Doppler shift
at C-band to the wind field and the radar configuration was then determined as shown
in Fig. 13.1 (Mouche et al. 2012). The spread indicated by the vertical bars can be
explained by several factors such as the shift in time between ECMWF outputs and
ASAR acquisition, errors in the wind from the model, other geophysical phenomena
such as rain
2 , errors in the Doppler estimates and the non-geophysical corrections
applied to the Doppler centroid. Using the CDOP relationship the residual Doppler
anomalies associated with the surface current can thus be determined with an r.m.s.
error of 5 Hz (equivalent to 0.2 m/s) at a spatial resolution of 8 km (along-track) by
4 km (across-track) as demonstrated by Johannessen et al. (2008) and Hansen et al.
(2011b).
The ascending Envisat tracks yield a more favorable imaging geometry with
respect to the core flow direction of the greater Agulhas Current as shown in
Fig. 13.2. Accordingly, the maximum mean radial Doppler velocity reaches about
2 Rain impacts the roughness and the local wind speed at a scale not resolved by Numerical Weather
Prediction (NWP) models.