Validation of Doppler Scatterometer Concepts using

Measurements from
the Black Sea Research Platform
Yury Yu. Yurovsky Vladimir N. Kudryavtsev Semyon A. Grodsky Bertrand Chapron
Remote Sensing Dept. Satellite Oceanography Lab. Dept. Atmos. Ocean. Sci. PDG-ODE-LOPS-SIAM
FSBSI MHI RAS RSHU University of Maryland IFREMER
Sevastopol, Russia Saint-Petersburg, Russia College Park MD, USA Plouzane, France
[email protected] [email protected] [email protected] [email protected]

Abstract—Doppler scatterometer is a prospective tool for sea (SKIM [6]) measurements. There is no well-established design
surface current measurements from space. Microwave Doppler of a Doppler scatterometer, since different observation strategy
signal is sensitive to various surface motions, including surface (e.g. incidence angle of 56◦ for WaCM/DopplerScatt or 6/12◦
wave orbital velocities and scatterer motions. In order to retrieve
the surface current form Doppler measurements, one needs for SKIM) leads to different drawbacks and benefits [13].
to understand the contributions of mean currents as well as However, a common line of the present concepts is the use
other factors. This work presents Ka-band dual-co-polarized of the highest possible radar operating frequency. Given a fixed
Doppler radar measurements in well-controlled field conditions antenna size, the Doppler velocity measurement accuracy is
carried out from the Black Sea research platform. Supplementary proportional to the radar wavelength. Up to a few millimeter
instruments include standard hydrometeorological sensors (wind
anemometer, air/water sensors, air humidity, pressure), and wavelength, the atmosphere is still transparent for electromag-
custom six-wire wave gauge. Sea surface currents are estimated netic waves, thus the Ka-band is the apparent choice for the
from video data captured by commercial video camera using instrument design [13].
wave dispersion analysis. The experiment strategy is to have At the same time, the Ka-band sea surface backscattering,
simultaneous radar, wind, wave, and current measurements that especially Doppler shifts, are not well studied because of pre-
can be used to decompose the mean Doppler signal into a set of
corresponding components. These measurements are to be used vious focus on the more traditional L/C/X/Ku-bands utilized
to establish an empirical model for the mean Doppler centroid. in scatterometers and synthetic aperture radars (SAR). Particu-
In turn, this may help to understand the relative contributions of larly, the first geophysical model function for the Doppler shift,
various factors to the Doppler centroid, including surface waves, the CDOP, was proposed for the C-band [14] based on the
and thus better configure future satellite Doppler scatterometer ENVISAT Advanced SAR data over the global ocean. Several
instrument.
Index Terms—Radar, backscattering, Doppler, velocity, sur- laboratory experiments [15]–[19] and a few recent airborne
face, current, wave, bias, dispersion studies [9], [20], were performed in the Ka-band. But, they
span only a small range of sea states and look geometries.
I. I NTRODUCTION Field experiments from static platforms [21]–[23] offer an
alternative way to better understand the backscattering physics.
Microwave Doppler scatterometer, a new tool for the fine- In this case the radar footprints can be designed to be smaller
resolution sea surface current vector measurement originating than the dominant wavelength providing direct measurements
from single line-of-sight projection retrievals from along-track of Doppler signal modulations by surface waves. For stationary
interferometry [1], [2] and Doppler anomaly [3], [4], is now and homogeneous processes, the temporal averaging of such
conceptualized in frames of several projects [5]–[9]. time-resolving measurements is equivalent to space averaging
The principal issue in the Doppler data inversion is the by satellite instrument.
fact that the Doppler radar measures the motion of surface Since 2009, such experiments have been conducted by a
waves advected by the surface current rather than the current dual-polarized Ka-band radar from the Black sea research
itself. Thus, similarly to altimeter measurements [10]–[12], platform [24], [25]. Here these data area used for validation of
the Doppler shift includes the sea state bias which should the Doppler scatterometer concept. The purpose of this note
be accounted for in Doppler data processing. Because of this is to present the measurement methodology and provide an
reason, the Doppler measurements are to be accompanied overview of the collected data.
by simultaneous wind (WaCM/DopplerScatt [9]) and/or wave
II. E XPERIMENTS AND E QUIPMENT
The research is sponsored by the Russian Science Foundation grant 17-77-
10052. The experiments are conducted by the FSBSI MHI within State Order The Black sea oceanographic research platform (Fig. 1) is
No. 0827-2018-0003. operated by the Black Sea Hydrophysical Polygon of Russian

978-1-5386-9527-2/18/$31.00 ©2018 IEEE

 Fig. 1. (a) Black sea research platform with installed (b) wave gauge, (c) wind sensors, and (d) Ka-band radar.

Academy of Science (RAS). The platform is moored 600 B. Hydro-Meteorological Sensors

m offshore in 30 m deep water (44◦ 230 3500 N, 33◦ 590 0400 E).
At onshore wind directions (easterly to south-westerly), the Supplementary instruments include standard meteorologi-
typical surface wavelength is ∼ 30 − 40 m at 10 − 15 m/s cal sensors installed at 23-m height mast (Fig. 1c): wind
winds, which ensures applicability of deep water conditions. anemometer, atmosphere temperature, humidity, and pressure.
At these winds, the wave state at the platform location can Water temperature sensor is submerged at 3m-depth. The data
be characterized as well-developed non-tidal deep water sea from meteo sensors are used to estimate wind speed at the
without strong swell. At offshore (northerly) winds, the fetch standard 10m-height using the COARE 3.0 algorithm [26].
is limited to 1 km, but wind speed can reach 40 m/s. Local Currents at 10-m depth are measured by submerged propeller
currents typically vary from 0 to 0.3 − 0.5 m/s, but may sensor.
accelerate ocassionally up to 1 m/s. Surface waves are measured by the six-wire wave gauge
(Fig. 1b). The wires are fixed at the center and vertices
of pentagon with 25-cm length edges. Directional frequency
A. Radar elevation spectra are estimated using the Extended Maximum
Likelyhood Method implemented in the DIWASP package
A Ka-band continuous-wave dual-copolarized Doppler scat- [27].
terometer is used for the radar measurements (Fig. 1d). It
consists of two co-aligned transmit/receive horn antennas
separated by 22 cm. The radar operates at hybrid polarization III. S EA S URFACE C URRENT M EASUREMENTS
mode transmitting wave of mixed polarization and receiving
two orthogonal (vertical and horizontal) modes of the received Doppler scatterometers are targeted at the retrieval of the
wave. This configuration implies a contamination of the co- surface currents to which they are sensitive, thus the ground
polarized signals (VV and HH) by cross-polarized signal (VH truth data for these currents are of special interest for this
and HV), which is very weak and can be disregarded (see work. From this reason we extend the standard 10-m depth
Appendix B in [24]). propeller current measurements with surface currents esti-
Detected Doppler signals (I/Q pair per polarization) are col- mated from video records of the sea.
lected by analog-to-digital (ADC) converter at 40 kHz/channel The video data were recorded by a digital video camera
sampling frequency. Optionally this ADC unit can be used installed atop of the radar and looking at its footprint (inci-
for simultaneous acquisition of six other channels (e.g. wave dence angle 0 − 70◦ ). The camera lens viewing angles are
gauge, video camera, wind sensor) to synchronize them with 28◦ /47◦ in vertical/horizontal planes. Output image sequence
radar data. is compressed by the MPEG2 codec in 1080i interlaced
Radar installation allows to vary the incidence angle, θ, from FullHD format (25 Hz frame rate, 1440 by 1080 pixels frame
0◦ to 70◦ with various radar-to-wind azimuths, but the pure size). Camera calibration shows no strong lens aberrations so
down-wind azimuth due to platform shadowing. At low/high simple projective transform is used to obtain georeferenced
incidence angles (θ is less/more than 45◦ ) the radar is placed images from known look geometry. Two approaches are used
on the top/bottom deck at 12/6-m height, respectively. to retrieve the sea surface currents.
 direction. Fortunately, shorter waves are more isotropic and
almost all azimuths can be detected from the dispersion curve
(Fig. 3c-e), thus providing a good constraint for the current
vector.
On the other hand, this method is based on the assumption
of linear relation between pixel brightness and surface slope,
that is valid only for small slopes and uniform sky brightness.
In the real world, the non-linearity of brightness-slope transfer
function results in high-order harmonics. This is seen in Fig.
3a-e as bright areas within the dispersion curve. Such situation
is typical for sun glint contamination.
For our purposes we use alternative approach. Each of the
3D-spectrum frequency slices is correlated with a sample slice
matrix, Q, corresponding to a given surface current (Fig. 4)
( √
Fig. 2. Scatterplot of detected particle direction and velocity colored by their 1, if ω = gk − k · V.
lifetime. Q(ω, V, k) = (2)
0, otherwise.

A. PIV Techniquie The surface current, V, is then varied within expected limits
−1 < Vx < 1 m/s and −1 < Vy < 1 m/s with 0.05 m/s step.
First, we applied the particle image velocimetry (PIV) for Correlation maps,
the sea surface images. The particles on the sea surface are R
standalone and groups of bubbles produced by breaking waves S(ω, k)Q(ω, V, k)dk
C(ω, V) = R R , (3)
and small floating sea debris. To separate the PIV particles S(ω, k)dk Q(ω, V, k)dk
from other bright features (sun glints, whitecaps), only small
shown in Fig. 3f-j visualize the surface current as a peak
(< 100 pixels2 ) and long-live events are selected. We found
located at {Vx , Vy }-point. We determine the peak location (the
that surface current speed estimates from PIV particles lasting
current estimate) and its width (the estimate error) through
longer than 8 sec converge and give an estimate of the drift
the first and second momenta of the correlation map at given
velocity (Fig. 2).
frequency, C(ω, V).
B. Dispersion Relation Analysis The peak width decreases with increasing wave frequency,
Second approach is similar to nautical radar [28], [29] and ω, because dispersion curve is better resolved at higher fre-
video [30], [31] data processing. Here we use the gravity wave quencies (Fig. 3f,g,h). With increase of ω, the dispersion curve
dispersion relationship to detect the current. The method is weakens while high-order harmonics become stronger. This
based on the analysis of the dispersion relation which, for results in a secondary peak in the correlation map, C(ω, V),
deep water waves, is at higher frequencies (Fig. 3i,j).
p Normally, the peak width (the estimate error) is either a
ω = gk − k · V, (1) monotonic function of frequency (if only the fundamental
where ω is the wave frequency, k is the wave vector, g is the mode is present) or has a minimum due to increasing high-
gravity acceleration, V is the current vector, the sign in the frequency tail (if harmonics are present). This property is
right hand side depends on the direction conventions (we use utilized as a criterion for the automated search of the sea
meteorological convention for wind, wave, and current, i.e. the surface current peak which is performed via analysis of
direction from which it originates). cumulative product of correlation maps at different discrete
The so-called dispersion shell is visualized in wavenumber- frequencies, ωi ,
frequency 3D-spectra, S(ω, k), of the video image brightness p
P (ωi , V) = C(ωi−1 , V)C(ωi , V), (4)
estimated using 3D Fast Fourier Transform of the georefer-
enced image sequence. Typical frequency slices of 3D video rather than just correlation maps at single frequency. This
spectra are shown in the top row of Fig. 3. In absence of makes the peak more “distinct” (compare middle and bottom
currents, each slice of the spectrum at a given frequency, ω, rows of Fig.3). The search for solution starts from the lowest
represents spectral densities distributed over a circle with the frequency and either proceeds until the highest frequency,
radius, ω 2 /g. If the current is present, the circle is distorted if peak width decreases with frequency, or stops, if peak
in accordance with (1) as seen in Fig. 3. Hence, the strength width reaches minimum. The frequency range is chosen from
of this distortion can be used to retrieve the current (1). 0.5 Hz to 1.5 Hz. In this frequency range, the dispersion
One of the limitation of the method is its dependence on curve is well resolved without significant distortions. Since
the wave azimuth spreading which determines the angular the peak width at fixed ω produced by the fundamental wave
distribution. Particularly, a single unidirectional wave can be mode is generally determined by the wavenumber spectrum
used to estimate only the current projection on the wave resolution (inverse to georeferenced image size) there is a
Fig. 3. (Top row) Slices of the 3D sea surface brightness spectrum at different frequencies computed for 15-min video fragment. Dispersive curve in absence
of currents is shown by black-white dashed line. (Middle row) Maps of correlation between the slices shown in the top row and sample slice matrix, Q.
(Bottom row) Cumulative product of correlation maps, P (ωi , V), e.g. n-th subplot in the bottom row is a product of first n subplots of the middle row, see
eq.(4). Camera height is 12 m, incidence angle is 48◦ , wind speed is 7 m/s.

following [32], the effective measurement depth is λ/(4π). In

our experiments we deal with currents integrated over depths
shallower than 0.5 m where we do not expect strong shears.
C. Comparison of the Methods
The 10-min average currents estimated from the wave
dispersion analysis are compared to the PIV currents and
the reference 10m-depth currents from submerged propeller
current meter (Fig. 5). Because the propeller measurements
are not available for the entire video data set, some points in
the top and bottom rows of Fig. 5 may not represent purely
synchronous data.
The current speed estimated by the two methods is well
correlated with the propeller current with differences of oppose
sign indicating that PIV current speed is higher than that
inferred from wave dispersion, which in turn is higher than
Fig. 4. Sample frequency slice, Q, of the 3D-spectrum at f = 1Hz for
V = 0.4 m/s current directed against x-axis. Overlaid white lines indicate
10m-depth current speed (Fig. 5a,c). The current direction for
theoretical dispersion curves for the current of the same direction but speed the PIV and the dispersion techniques are almost identical
varied from 0 to 0.5 m/s. (Fig. 5b), while there is current rotation in the upper 10m
layer (135◦ for three samples, Fig. 5d).
An explanation for the vertical differences in current speed
natural threshold for the peak width that separates successful and direction is the presence of wind-induced surface drift. If it
and unsuccessful algorithm outputs. is accounted for from observed wind speed, αU , the corrected
Note that surface current peaks may differ for different wave differences in speed decrease while direction difference for
frequencies because of the vertical current shear. However, the three “abnormal” cases decrease significantly (red dots in
Fig. 5. Scatterplots for the current estimated from wave dispersion analysis versus (a,b) PIV-derived sea surface currents, and (c,d) propeller sensor currents
measured at 10-m depth. (a,c) – current magnitude, (b,d) – current direction. Blue – measurements, red – measurements including estimated wind-drift current.

the Fig. 5). The value of α = −0.75% for PIV-to-dispersion The data set is refined by filtering out short-fetch sea states
difference suggests that surface markers, such as bubbles and caused by offshore winds and cases with different wind and
foam, slip with respect to the sea surface due to their wind wave directions (> 45◦ ).
drag. For the propeller-to-dispersion comparison the wind drift
A. Wave-Induced Doppler Centroid Contribution
coefficient α = 1.65% is about twice less than its classical
3%U estimate, but in line with even smaller shear α = 1.3% First, we compare the measured DC with the surface current
independently measured near to the platform [33]. projection on the radar line-of-sight, Vd sin θ cos φd , where
For radar data analysis, the currents derived from wave φd is the radar-to-current azimuth (Fig. 7). This comparison
dispersion analysis are used by default, Vd . If this estimate directly demonstrates that the measured radar DC is not the
is unavailable (e.g. too many sun glints), we reconstruct the surface current. Although points grouped by the incidence
current using the vertical shear, αU , either from PIV or angle correlate (better for VV than for HH), there is an
propeller data. The latter data are used if video records are apparent deviation between radar data and in-situ currents.
not available. Following previous study [2], [3], this difference results from
the correlation between local Doppler velocity (wave orbital
IV. R ESULTS motion) and local radar cross-section, as well as, in terms
Raw radar records, totalling ∼ 60 hours, are partitioned into of two-scale separation, from the motion of scatterers (Bragg
5-min samples. Doppler centroids (DC) are calculated via the wave or specular point velocity). This is highlighted by Fig.
1st moment of the Doppler spectrum of 5-min fragment (pos- 8 showing the dependence of DC on wind speed, the main
itive Doppler frequency corresponds to approaching target). factor determining the sea state. The DC is apparently higher
For each fragment, the wind speed and wave spectrum are than the wind drift even if it was estimated as 3%U .
estimated, while the surface currents are available only for This shows that the difference between DC and surface
∼ 50% of radar samples. The gross statistics for 2009-2017 current is related to the impact of radar-wave correlations
dataset is shown in the Fig. 6. The moderate to high incidence rather than wind-induced vertical shear. The wave-induced DC
angles, θ > 20◦ , are covered rather uniformly for wind speeds contribution is usually parameterized in terms of a Modulation
U < 12 − 15 m/s, while at low incidence angles, θ < 20◦ , Transfer Function (MTF) [34], [35] or via a gain factor (G-
only calm sea states, U < 7 m/s, are captured. With respect to factor) introduced in [3], [6]. As seen in Fig. 8, the wave-
the radar-to-wave azimuth, φ (zero corresponds to the upwave induced bias depends on incidence angle, marked by color,
direction), the data are mostly concentrated around 120◦ at indicating variability of the MTF with θ. Generally, this
small θ < 10◦ . At higher θ, the coverage is more uniform. variability is more complex because the MTF depends on the
Fig. 6. Number of 5-min samples for 2009-2017 experiments as function of incidence angle, θ, and (a) wind speed, (b) significant wave height (SWH), (c)
peak wave period, and (d) radar-to-wave azimuth (zero is upwave).

Fig. 7. Doppler centroids versus line-of-sight sea surface current projection, Vd sin θ cos φd , for (a) VV polarization and (b) HH polarization.

shape of function relating cross-section and incidence angle, we compare our DC to the published data mentioned above,
which in turn depends on the wind speed and azimuth (tilt both rescaled from Hz to m/s for consistency. Note, the Bragg
modulation), and small scale roughness modulation by longer wave phase velocity for C/X/Ku/Ka-band is 24/23/27/35 cm/s
waves (hydro-modulation). More details on analysis of the thus the major part of the differences between the bands, if
MTF for the present data set can be found in [25], [36]. any, will origin from MTF differences.
The DC as function of incidence angle, θ, is presented in
B. Comparison with Published Data the Fig. 9 for available wind/wave conditions. It monotonically
Available measurements of Ka-band DC in a wide range of grows from zero at nadir θ = 0◦ to maximal observed values
incidence angles and sea states are rare. To the best of our of 0.5 − 0.9 m/s for VV polarization and 0.7 − 1.2 m/s for
knowledge, there are only C-band Doppler GMF [14] derived HH polarization at θ = 70◦ in the upwind direction. In the
from the Envisat ASAR data at θ = 17 − 42◦ (VV and HH downwind direction, the DC saturates for θ >≈ 30◦ at 0.5/0.6
polarizations) and recent Ka-band DopplerScatt model [9] for m/s for VV/HH polarization, respectively. Both upwind and
θ = 56◦ (VV polarization only). Some data are published in downwind DCs are significantly higher than the pure Bragg
[20] for Ka-band AirSWOT flights at 2◦ < θ < 20◦ and in phase velocity, again indicating the influence of wave-induced
[16] for Ku-band FPN field measurements. With our data we Doppler shift. In the crosswind direction, the DC is close zero,
can now take a look at the dual-copolarized DC behaviour as except at large θ = 70◦ for which it becomes positive (motion
a whole, at least in the Ka-band. For the sake of curiosity, towards an observer).
Fig. 8. Doppler centroids versus incidence-plane wind projection, U cos φ, for (a) VV polarization and (b) HH polarization. Black line indicates the upper
limit for the expected wind drift velocity, 3%U .

The C-band CDOP in general is in good agreement with our is important for VV polarization and is dominating for HH
Ka-band data for HH polarization, while for VV polarization polarization. Recently, we reported that Doppler shift asso-
there are some discrepancy at θ > 30◦ , especially in downwind ciated with breaking wave is not as high as breaking wave
direction where CDOP model exhibits a peak in magnitude phase velocity, but are closer to its orbital velocity [25], [37].
while our data do not. The Ka-band DopplerScatt DCs are also For ideally co-aligned cross-look wind, waves, and surface
in a good consistence with our data on average with somewhat current their corresponding line-of-sight components are all
smaller variance in upwind direction. The three flights Ka- zero (except for the above mentioned hydro-wave-induced
band AirSWOT data (Fig. 16c in [20]) is also higher than our bias). The same holds for breakers travelling directly in the
DCs, probably due to a fact that the AirSWOT data are for 8 cross-look direction. However, waves that break are shorter
m/s and 40 m peak wind wave length, while our present data than peak waves [38] and thus they have a wider directional
at θ < 20◦ corresponds mainly to weaker U = 5 − 6 m/s and spreading and an observer looking crosswind will still see
less-developed wind waves. The FPN Ku-band measurements some approaching and moving away breakers with Doppler
(Figs. 1,2 in [16], DC are estimated numerically from the shifts of opposite signs. At the same time, our observations
published Doppler spectra) at U = 7 − 8 m/s in upwind [24] indicate strong anisotropy in the sea surface cross-section
direction and VV/HH polarizations coincide with our data, at large θ suggesting that breakers, vastly responsible for the
as well as with DopplerScatt GMF, with somewhat increased backscatter in this regime, are essentially anisotropic scatter-
level at θ = 45◦ . The crosswind DC increase at θ = 70◦ is ers. In turn, this means that approaching breakers contribute
also confirmed by the FPN data with the same higher DC level much stronger than the moving away ones, and thus may
for HH polarization. Specific crosswind DC behaviour at large explain the observed positive crosswind DC.
θ can be explained by the following mechanism.
First, to the first order in wave slopes, there is no tilt V. S UMMARY
modulation when a radar looks along wave crests. The Doppler This message is aimed at presenting radar Doppler measure-
signal has opposite sign extrema on front and rear wave slopes, ments collected during multi-year experiments from the Black
but is weighted by corresponding front and rear slope cross- sea research platform. The measurements performed in the Ka-
section. Thus, the non-zero cross-wind DC can result from band are still quite rare in literature, thus these data can be
differences in radar backscattering from front and rear wave used for validations of future Ka-band Doppler scatterometer
slopes due to the hydrodynamic modulation of small-scale concepts such as (SKIM [6], WaCM/DopplerScatt [9]).
roughness. Particularly, the positive sign of cross-wind DC The main goal of the Doppler scatterometry is the surface
at large θ means that front slopes have larger backscattering current retrieval. That is why we pay special attention to
than rear slopes. independent sea surface current measurements. A practical
Second, at large θ, the backscattering from breaking waves method for the sea surface current estimation is proposed
Fig. 9. Doppler centroids versus incidence angle, θ, (a) VV polarization and (b) HH polarization. All covered sea states, mostly U = 4 − 12 m/s, see Fig. 6
for details. Artificial Gaussian noise is added to θ to improve visibility of the plots.

based on the analysis of conventional video camera records. breaker is faced towards the radar, while moving away is not,
The georeferenced video images are very similar to the PPI- thus the observed bias is positive.
images of nautical radars for which surface current and shallow Overall, the reported Ka-band data are not in conflict with
sea depth estimation is developed based on the wave dispersion any other published measurements in the C/Ku/Ka-bands [9],
relationship analysis [28], [29]. Similarly to this approach, we [14], [16], [20]. These data represent a wide range of incidence
compute frequency-wavenumber 3D-spectrum of video image angles, azimuths, and sea states. The ongoing work focuses on
sequence in order to analyze surface wave dispersion relation-
√ filling gaps in the data set with new measurements and building
ship and retrieve the current from its deviation from ω = gk. an empirical Ka-band DC model.
To avoid influences of high-order brightness harmonics we
correlate the frequency spectrum slices with expected sample R EFERENCES
slices for variable currents. The correlation maps provide [1] R. M. Goldstein and H. A. Zebker, “Interferometric radar measurement
detectable peaks corresponding to the sea surface current. The of ocean surface currents,” Nature, vol. 328, pp. 707–709, Aug. 1987.
validation of the method is based on the comparison with [2] R. Romeiser and D. R. Thompson, “Numerical study on the along-track
interferometric radar imaging mechanism of oceanic surface currents,”
PIV processing of the same video data and submerged (10- IEEE Trans. Geosci. Remote Sens., vol. 38, no. 1, pp. 446–458, Jan
m) propeller current meter data. 2000.
[3] B. Chapron, F. Collard, and F. Ardhuin, “Direct measurements of ocean
Resulting sea surface currents are compared to the radar surface velocity from space: Interpretation and validation,” J. Geophys.
Doppler centroids (DC) demonstrating, first, an ability of Res. (Oceans), vol. 110, p. 7008, 2005.
[4] H. Johnsen, V. Nilsen, G. Engen, A. A. Mouche, and F. Collard, “Ocean
Doppler scatterometer to measure the current, and, second, doppler anomaly and ocean surface current from sentinel 1 tops mode,”
the contribution of the so-called wave-induced bias originat- in Proc. Int. Geosci. Remote Sens. Symp., July 2016, pp. 3993–3996.
ing from a correlation between wave orbital motions and [5] M. A. Bourassa, E. Rodriguez, and D. Chelton, “Winds and currents
mission: Ability to observe mesoscale air/sea coupling,” in Proc. Int.
cross-sections. Here we give a very cursory overview of the Geosci. Remote Sens. Symp., July 2016, pp. 7392–7395.
measured DC as a function of incidence angle and radar-to- [6] F. Ardhuin, Y. Aksenov, A. Benetazzo, L. Bertino, P. Brandt, E. Caubet,
wind(wave) azimuth. We note an interesting feature of the B. Chapron, F. Collard, S. Cravatte, F. Dias, G. Dibarboure, L. Gaultier,
J. Johannessen, A. Korosov, G. Manucharyan, D. Menemenlis,
positive crosswind bias at large incidence angles, θ = 70◦ M. Menendez, G. Monnier, A. Mouche, F. Nouguier, G. Nurser,
confirmed by previously reported Ku-band Doppler spectra by P. Rampal, A. Reniers, E. Rodriguez, J. Stopa, C. Tison, M. Tissier,
Plant [16]. This is qualitatively explained by the contribution C. Ubelmann, E. van Sebille, J. Vialard, and J. Xie, “Measuring
currents, ice drift, and waves from space: the Sea Surface KInematics
of breaking waves having wide azimuth spreading but narrow Multiscale monitoring (SKIM) concept,” Ocean Science Discussions,
backscattering radiation pattern (anisotropic scatterer). Thus vol. 2017, pp. 1–26, 2017. [Online]. Available: https://www.ocean-sci-
breakers running at the same speed symmetrically to the left discuss.net/os-2017-65/
[7] Q. Bao, M. Lin, Y. Zhang, X. Dong, S. Lang, and P. Gong, “Ocean
and to the right from the dominant wave (cross-look) direction surface current inversion method for a doppler scatterometer,” IEEE
contribute differently to the DC. Particularly, the approaching Trans. Geosci. Remote Sens., vol. 55, no. 11, pp. 6505–6516, Nov 2017.
 [8] C. Gommenginger, B. Chapron, A. Martin, J. Marquez, C. Brownsword, ality and surface currents.” J. Geophys. Res., vol. 90, no. C1, pp. 1049–
and C. Buck, “SEASTAR: A new mission for high-resolution imaging 1059, 1985.
of ocean surface current and wind vectors from space,” in Proceedings [29] C. M. Senet, J. Seemann, and F. Ziemer, “The near-surface current
of the European Conference on Synthetic Aperture Radar, EUSAR, vol. velocity determined from image sequences of the sea surface,” IEEE
June, 2018, pp. 1433–1436. Trans. Geosci. Remote Sens., vol. 39, no. 3, pp. 492–505, 2001.
[9] E. Rodriguez, A. Wineteer, D. Perkovic-Martin, T. Gál, B. Stiles, [30] J. P. Dugan, H. Suzukawa, C. P. Forsyth, and M. S. Farber, “Ocean wave
N. Niamsuwan, and R. Rodriguez Monje, “Estimating ocean vector dispersion surface measured with airborne ir imaging system,” IEEE
winds and currents using a Ka-band pencil-beam doppler scatterometer,” Trans. Geosci. Remote Sens., vol. 34, no. 5, pp. 1282–1284, 1996.
Remote Sens., vol. 10(4), April 2018, 576. [Online]. Available: [31] N. J. M. Laxague, B. K. Haus, D. G. Ortiz-Suslow, C. J. Smith,
http://www.mdpi.com/2072-4292/10/4/576 G. Novelli, H. Dai, T. Ozgokmen, and H. C. Graber, “Passive optical
[10] M. A. Srokosz, “On the joint distribution of surface elevation and slopes sensing of the near-surface wind-driven current profile,” Journal of
for a nonlinear random sea, with an application to radar altimetry,” J. Atmospheric and Oceanic Technology, vol. 34, no. 5, pp. 1097–1111,
Geophys. Res. (Oceans), vol. 91, no. C1, pp. 995–1006, 1986. 2017.
[11] E. Rodriguez, Y. Kim, and J. M. Martin, “The effect of small-wave [32] R. H. Stewart and J. W. Joy, “HF radio measurements of ocean surface
modulation on the electromagnetic bias,” J. Geophys. Res. (Oceans), currents,” Deep Sea Res., vol. 21, pp. 1039–1049, 1974.
vol. 97, no. C2, pp. 2379–2389, 1992. [33] V. V. Malinovsky, V. A. Dulov, A. E. Korinenko, A. N. Bol’Shakov, and
[12] T. Elfouhaily, D. R. Thompson, B. Chapron, and D. Vandemark, V. E. Smolov, “Field investigations of the drift of artificial thin films on
“Improved electromagnetic bias theory: Inclusion of hydrodynamic the sea surface,” Izvestia Atmosph. Ocean. Phys., vol. 43, pp. 103–111,
modulations,” J. Geophys. Res. (Oceans), vol. 106, no. C3, pp. 4655– 2007.
4664, 2001. [34] W. C. Keller and J. W. Wright, “Microwave scattering and the straining
[13] E. Rodriguez, “On the optimal design of doppler scatterometers,” of wind-generated waves,” Radio Sci., vol. 10, pp. 139–147, 1975.
Remote Sens., vol. 10(11), November 2018, 1765. [Online]. Available: [35] W. J. Plant, “The Modulation Transfer Function: Concept and Appli-
https://www.mdpi.com/2072-4292/10/11/1765 cations,” Radar Scattering from Modulated Wind Waves, pp. 155–172,
[14] A. A. Mouche, F. Collard, B. Chapron, K. F. Dagestad, G. Guitton, J. A. 1989.
Johannessen, V. Kerbaol, and M. W. Hansen, “On the Use of Doppler [36] Yu. Yu. Yurovsky, S. A. Grodsky, V. N. Kudryavtsev, and B. Chapron,
Shift for Sea Surface Wind Retrieval From SAR,” IEEE Trans. Geosci. “Wave-induced Doppler shift of Ka-band radar signal backscattered
Remote Sens., vol. 50, no. 7, pp. 2901–2909, July 2012. from the sea surface,” in 2017 Progress in Electromagnetics Research
[15] J.-P. Giovanangeli, L. F. Bliven, and O. Le Calve, “A wind-wave tank Symposium - Fall (PIERS - FALL), Nov 2017, pp. 2299–2306.
study of the azimuthal response of a Ka-band scatterometer,” IEEE [37] Yu. Yu. Yurovsky, V. N. Kudryavtsev, B. Chapron, and S. A. Grodsky,
Trans. Geosci. Remote Sens., vol. 29, pp. 143–148, Jan. 1991. “How Fast are Fast Scatterers Associated with Breaking Wind Waves?”
[16] W. J. Plant, “A model for microwave Doppler sea return at high inci- in Proc. Int. Geosci. Remote Sens. Symp., Valencia, Spain, 22-27, July
dence angles: Bragg scattering from bound, tilted waves,” J. Geophys. 2018, pp. 142–145.
Res. (Oceans), vol. 102, pp. 21 131–21 146, 1997. [38] W. J. Plant, “Whitecaps in deep water,” Geophys. Res. Lett.,
[17] S. A. Ermakov, I. A. Kapustin, and I. A. Sergievskaya, “Tank study of vol. 39, no. 16, pp. n/a–n/a, 2012, l16601. [Online]. Available:
radar backscattering from strongly nonlinear water waves,” Bulletin of http://dx.doi.org/10.1029/2012GL052732
the Russian Academy of Science, Phys., vol. 74, pp. 1695–1698, 2010.
[18] S. A. Ermakov, I. A. Kapustin, V. N. Kudryavtsev, I. A. Sergievskaya,
O. V. Shomina, B. Chapron, and Yu. Yu. Yurovskiy, “On the doppler
frequency shifts of radar signals backscattered from the sea surface,”
Radiophysics and Quantum Electronics, vol. 57, no. 4, pp. 239–250,
2014.
[19] O. Boisot, L. Amarouche, J. C. Lalaurie, and C. A. Guerin, “Dynamical
Properties of Sea Surface Microwave Backscatter at Low-Incidence:
Correlation Time and Doppler Shift,” IEEE Trans. Geosci. Remote Sens.,
vol. 54, no. 12, pp. 7385–7395, Dec 2016.
[20] F. Nouguier, B. Chapron, F. Collard, A. Mouche, N. Rascle, F. Ardhuin,
and X. Wu, “Sea surface kinematics from near-nadir radar measure-
ment,” IEEE Trans. Geosci. Remote Sens., 2018, accepted.
[21] F. Feindt, J. Schroeter, and W. Alpers, “Measurement of the ocean wave-
radar modulation transfer function at 35 GHz from a sea-based platform
in the North Sea,” J. Geophys. Res. (Oceans), vol. 91, no. C8, pp. 9701–
9708, 1986.
[22] W. J. Plant and W. Alpers, “An Introduction to SAXON-FPN,” J.
Geophys. Res. (Oceans), vol. 99, no. C5, pp. 9699–9703, 1994.
[23] W. C. Keller, W. J. Plant, R. A. Petitt, and E. A. Terray, “Microwave
backscatter from the sea: Modulation of received power and Doppler
bandwidth by long waves,” J. Geophys. Res. (Oceans), vol. 99, no. C5,
pp. 9751–9766, 1994.
[24] Yu. Yu. Yurovsky, V. N. Kudryavtsev, S. A. Grodsky, and B. Chapron,
“Ka-Band Dual Copolarized Empirical Model for the Sea Surface Radar
Cross Section,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 3, pp.
1629–1647, 2017.
[25] Yu. Yu. Yurovsky, V. N. Kudryavtsev, B. Chapron, and S. A. Grodsky,
“Modulation of Ka-band Doppler Radar Signals Backscattered from the
Sea Surface,” IEEE Trans. Geosci. Remote Sens., vol. 56, no. 5, pp.
2931–2948, May 2018.
[26] C. W. Fairall, E. F. Bradley, J. E. Hare, A. A. Grachev, and J. B. Edson,
“Bulk Parameterization of Air Sea Fluxes: Updates and Verification for
the COARE Algorithm,” Journal of Climate, vol. 16, pp. 571–591, 2003.
[27] D. Johnson, “DIWASP, a directional wave spectra toolbox for MATLAB:
User Manual,” Research Report WP-1601-DJ (V1.1), Centre for Water
Research, University of Western Australia, 2002.
[28] I. R. Young, W. Rosenthal, and F. Ziemer, “A three-dimensional analysis
of marine radar images for the determination of ocean wave direction-