1
Please note that this is an author-produced PDF of an article accepted for publication following peer review. The definitive
publisher-authenticated version is available on the publisher Web site.
Geoscience and Remote Sensing Symposium (IGARSS),
2016 IEEE International. 10-15 July 2016
Pages 3993-3996
http://dx.doi.org/10.1109/IGARSS.2016.7730038
http://archimer.ifremer.fr/doc/00356/46713/
© IEEE 2016
Achimer
http://archimer.ifremer.fr
Ocean doppler anomaly and ocean surface current from
Sentinel 1 tops mode
Johnsen Harald
1, *
, Nilsen Vegard
1
, Engen Geir
1
, Mouche Alexis
2
, Collard Fabrice
3
1
Northern Research Institute, Box 6434, N-9294 Tromsoe, Norway
2
Laboratoire d'Ocanographie Spatiale - Ifremer, 29280 Plouzane, France
3
OceanDataLab, 29280 Locmaria Plouzane, France
* Corresponding author : Harald Johnsen, email address : harald.johnsen@norut.no
Abstract :
Processing and analysis of Doppler information from Sentinel 1A Interferometric Wide (IW) and Extra
Wide (EW) modes are performed for assessing the capabilities of mapping ocean surface current field.
Data from Agulhas (South-Africa) and Norwegian Coast are used in combination with numerical models,
higher-order satellite products, and Lagrangian drifters. Results show strong Doppler signal and
dynamics from coastal areas caused by a mixture of surface current and wind/wave induced drifts at a
spatial resolution of around 2 km2 in IW mode and 4km2 in EW mode. Doppler values of up to 70 Hz
are observed, corresponding to a surface drift velocity of 3.5 m/s. The Sentinel 1 retrieved surface
current component is in reasonable agreement with the circulation models and drifter measurements.
Surface current values up to 1.5 m/s are observed in the central Agulhas current, with a standard
deviation of around 0.39 m/s with respect to Lagrangian drifters.
Keywords : Doppler effect, Sea surface, Sea measurements, Ocean temperature, Surface treatment,
Antennas
OCEAN DOPPLER ANOMALY AND OCEAN SURFACE CURRENT FROM SENTINEL 1 TOPS MODE
Harald Johnsen
1
, Vegard Nilsen
1
, Geir Engen
1
, Alex A. Mouche
2
, Fabrice Collard
3
1
Northern Research Institute, Box 6434, N-9294 Tromsoe, Norway
2
Laboratoire d’Océanographie Spatiale - Ifremer, 29280 Plouzane, France
3
OceanDataLab, 29280 Locmaria Plouzane, France
ABSTRACT
Processing and analysis of Doppler information from
Sentinel 1A Interferometric Wide (IW) and Extra Wide
(EW) modes are performed for assessing the capabilities of
mapping ocean surface current field. Data from Agulhas
(South-Africa) and Norwegian Coast are used in
combination with numerical models, higher-order satellite
products, and Lagrangian drifters.
Results show strong Doppler signal and dynamics from
coastal areas caused by a mixture of surface current and
wind/wave induced drifts at a spatial resolution of around 2
km
2
in IW mode and 4km
2
in EW mode. Doppler values of
up to 70 Hz are observed, corresponding to a surface drift
velocity of 3.5 m/s. The Sentinel 1 retrieved surface current
component is in reasonable agreement with the circulation
models and drifter measurements. Surface current values up
to 1.5 m/s are observed in the central Agulhas current, with
a standard deviation of around 0.39 m/s with respect to
Lagrangian drifters.
Index Terms— Sentinel 1A, Doppler Anomaly, Ocean
Surface Current
1. INTRODUCTION
The Doppler centroid anomaly recorded over ocean with a
Synthetic Aperture Radar (SAR) can be used to obtain range
directed velocity which has been demonstrated to provide
valuable estimates of the near surface wind speed, ocean
surface current [1],[2],[3] and sea ice sea drift [4]. The
Doppler centroid (Dc) frequency recorded over ocean
differs from the geometric and instrument Doppler centroids
arising from antenna miss pointing, satellite orbit and
attitude. This difference, called the Doppler centroid
anomaly (or geophysical Doppler), is a direct measure of the
line-of-sight (radial) velocity of the moving ocean surface
and is thus highly sensitive to the motion of the scatterers
induced by the surface currents, winds and ocean waves.
A high-precision Doppler centroid anomaly estimator
was developed and implemented as part of the Sentinel-1
Level 2 ocean processor [5],[6]. The basis for our analysis is
the Sentinel-1 Level 2 products from IW and EW modes
collocated with Lagrangian drifters and model data. A short
description of the data used is given in Section 2.
In Section 3 we first describe the procedure used for
calibrating the Doppler measurements to account for
instrumental and orbit effects. Secondly, the methodology
for retrieving the ocean surface current (OSC) from the
Doppler anomaly is outlined. In Section 4 some validation
results are presented and discussed.
2. DATA DESCRIPTION
The following data sources are used in the analysis:
• Sentinel 1 WV, IW and EW data (earth.esa.int)
• AROME local weather model (met.no)
• ECMWF global weather model (www.ecmwf.int)
• OSCAR circulation model (oscar.noaa.gov)
• GlobCurrent data product (www.globcurrent.org)
• Lagrangian drifters (www.metocean.com)
• Sea surface temperature product (www.ifremer.fr)
• Ocean Virtual Laboratory (ovl.oceandatalab.com)
The Sentinel 1 IW and EW data are processed into a Level 2
Doppler product at a resolution of 2 and 4 km
2
, respectively.
3. METHODOLOGY
The Doppler frequency estimated from the single-look
complex data consist in general of several terms:
(1)
ϖ
dc
=
ϖ
dc
phys
+
ϖ
dc
geo
+
ϖ
dc
elec
+Δ
ϖ
dc
where
ϖ
dc
phys
is the geophysical term,
ϖ
dc
geo
is the
geometric/attitude term,
ϖ
dc
elec
is the antenna electronic
miss pointing, and
Δ
ϖ
dc
is a residual error coming from
imperfect prediction of the non geophysical terms or other
unknown biases. In order to derive an estimate of the
geophysical Doppler ( ), a careful prediction and
removal of the non-geophysical contributions from Eq.(1)
are needed.
3.1. Doppler calibration
The geometric and the antenna electronic miss pointing
terms are initially compensated for within the Sentinel 1
Level 2 processor. However, it turns out that the geometric
ϖ
dc
phys
Doppler computed from the attitude described by the
quaternions does not predict the Dc estimated from the SAR
data. A significant latitude dependency is observed in the
estimated Dc, but not in the geometric Doppler (see Fig. 1).
Figure 1: Latitude dependency in Dc estimated from S1A WV
acquired of land areas. Data from February 2016. The error bars
are the variability of the Dc estimates. The dotted lines are the
corresponding geometric Doppler.
The electronic miss pointing Doppler predicted by the antenna
model describes reasonably well the relative variations as function
of elevation angles. However, the absolute Doppler value and the
jumps in Doppler overs swaths are not always well predicted. In
Figure 2 is shown the Dc estimates as function of incidence angles
for the IW swaths and the corresponding electronic miss pointing
Doppler computed from the antenna model. Similar results are
observed for the EW mode.
a) b)
Figure 2: a) Dc as function of incidence angle estimated from S1A
IW mode acquired over rain forest, b) corresponding Doppler
profiles estimated from antenna model.
A recalibration has been applied to the estimated Dc
values to compensate at best the observed deviations shown
in Figs. 1 and 2. The approach selected for our analysis is as
follows:
a. Remove the electronic miss pointing Doppler from
the observed Doppler, (
δϖ
≡
ϖ
dc
−
ϖ
dc
elec
) using the
antenna model
b. Make the residual Doppler,
δϖ
continuous from
one swath to another by merging the Doppler
values in azimuth at the swath borders
c. Use land areas within the scene to absolute
calibrate the residual Doppler, (
ϖ
dc
phys
≡
δϖ
−
δϖ
LAND
).
Here
δϖ
LAND
is the mean Doppler over land areas
within the swath.
The residual Doppler is then an estimate of the Doppler
anomaly or the geophysical Doppler (
ϖ
dc
phys
) of Eq.(1).
3.2. Doppler anomaly analysis
The Doppler anomaly contains information of the ocean
surface wind field, sea state and ocean surface current. In
order to retrieve the surface current, the contributions from
wind/wave induced motion must be predicted and removed.
The wind/wave Doppler contribution (
ϖ
dc
cdop
U
10
( )
) is
predicted using the CDOP [3] with input the best possible
wind field, U
10
extracted from numerical weather model. An
estimate of the ocean surface current (OSC) induced
Doppler is then given as:
(2)
ϖ
dc
osc
=
ϖ
dc
phys
−
ϖ
dc
cdop
Eq.(2) is then converted to ground range radial surface
current by the relation:
(3)
U
r
=
ϖ
dc
osc
2k
rad
sin
θ
where
k
rad
is the radar wavenumber and
θ
is the local
incidence angle. The
U
r
values are collocated with
Lagrangian drifters and circulation models for validation.
For the Norwegian Coastal areas, the wind vector field is
provided from the AROME model with a spatial resolution
of 2.5 km. For the Agulhas area, the global wind model
from ECMWF is used. For the Agulhas areas, the collocated
Oscar current field and the GlobCurrent field are extracted
using the Ocean Virtual Laboratory toolbox.
4. RESULTS
The analysis of Sentinel 1 Doppler anomaly shows strong
signatures and dynamic of surface current and wind/wave
induced drifts. In Fig. 3 is shown an example of S1A IW
ascending acquisition over the Agulhas current outside
South-Africa. The data are processed into a Level 2 product
with a spatial resolution of around 2 km
2
.
A strong geophysical Doppler signal (up to 70Hz) from
the Agulhas current is observed (red color) near the
coastline in Fig. 3a, while in the lower part of the image a
mixture of wind/wave induced Doppler signal is observed
(yellow/blue colour). A ground range velocity (Fig. 3b) of
up to 3.5 m/s is observed in the central current area. A
negative velocity means towards the radar look direction. A
Lagrangian drifter was within the area at the time of
acquisition as indicated in Fig. 3b (red arrow), with a
velocity of 1.1 m/s and direction of 228 degN. In this case
the wind speed and direction is around 9 m/s and 230 degN,
contributing significantly to the observed S1A velocity.
a)
b)
Figure 3: Sentinel 1A IW Doppler anomaly (a) and the ground
range velocity component (b) from ascending mode acquisition
over the Agulhas area outside South-Africa. The Lagrangian
drifter position and direction is indicated with a red arrow in (b).
The velocity from the drifter was 1.1 m/s.
A second test area is outside the North-Norwegian coast
where the continental shelf and the coastal current are close
to the coastline. The area is also exposed to low-pressure
systems from North-Atlantic with strong atmospheric fronts.
Is it expected that the Doppler anomaly is dominated by
wind/wave effects, and less by coastal current. Fig. 4 shows
a Sentinel 1A IW ascending acquisition from the coast of
North-Norway. The AROME model (Fig. 4c) together with
CDOP is used to predict and remove the wind/wave signal
from the S1A Doppler anomaly. The final ocean surface
current field (ground range component) is shown in Fig. 4c.
A coastal current signal component of around 0.4 m/s is
retrieved from the Doppler anomaly.
Altogether 32 scenes in IW and EW were analysed using
the procedure described above. Scatterplots of the surface
current speed (ground range component) from S1A versus
Oscar, GlobCurrent and Lagrangian drifter, are given in Fig.
5 and 6. No filtering is applied to the data. A standard
deviation of 0.39 m/s and a bias of 0.38 m/s are observed
between S1A and Lagrangian drifters. The main
contribution to the standard deviation and bias is expected to
be the uncertainty in attitude as well as in the wind field
used to remove the contribution to the Doppler anomaly
from wind and waves.
For future work, more data will be collected and the use
of S1 wind field itself will be exploited in the analysis. The
foreseen improved prediction of geometric and electronic
miss pointing Doppler will improve the Sentinel 1 Doppler
anomaly measurements.
a)
b)
c)
d)
Figure 4: Sentinel 1A IW intensity (a), ocean surface velocity
(ground range component) (b), AROME wind speed (c), and
retrieved ocean surface current (ground range component) (d).
Figure 5: Scatterplot of ocean surface velocity (radial
component) from S1A versus Oscar and GlobCurrent.
Figure 6: Scatterplot of ocean surface velocity (radial
component) from S1A versus Lagrangian drifter data.
Acknowledgement:
This work is performed under the S1 Mission Performance Centre
activities and the SEOM S1-4SCI Ocean Study, funded by
ESA/ESRIN, and CIRFA – 237906/O30 funded by Norwegian
Research Council.
REFERENCES
[1] B. Chapron, F. Collard, F. Ardhuin, “Direct measurements of
ocean surface velocity from space: Interpretation and validation”,
J. Geophysical Research, Vol. 110, No. C7, pp. 10.1029, 2005
[2] J. A. Johannessen, B. Chapron, F. Collard, V. Kudryavtsev, A.
Mouche, D. Akimov, K. F. Dagestad, “Direct ocean surface
velocity measurements from space: Improved quantitative
interpretation of Envisat ASAR observations”, Geophysical
Research Letters, Vol. 5, doi:10.1029/2008GL035709, 2008
[3] A. Mouche, F. Collard, B. Chapron, K. F. Dagestad, G.
Guittion, J. A. Johannessen, V. Kerbaol, M. W. Hansen, “On the
use of Doppler shift for sea surface wind retrieval from SAR”,
IEEE Trans. Geoscience and Remote Sensing, Vol. 50, pp.2901-
2909, doi:10.1109/TGRS.2011.2174998, 2012
[4] T. Kræmer T., H. Johnsen, C. Brekke, “Emulating Sentinel-1
Doppler Radial Ice Drift Measurements Using Envisat ASAR
Data”, IEEE Trans. Geoscience and Remote Sensing, Vol.12,
No.53, DOI: 10.1109/TGRS.2015.2439044, pp: 6407 – 6418,
August 2015
[5] G. Engen, H. Johnsen, “Sentinel-1 Doppler and Ocean Radial
Velocity Algorithm Definition”, ESA Doc. S1-TN-NRT-53-0658,
ISBN 978-82-7492-311-9, ISSN 1890-5226, v1.4, 2015
[6] G. Engen, H. Johnsen, Y. Larsen, “Sentinel-1 geophysical
Doppler product – performance and application”, Proc. of
EUSAR2014, Berlin, 3-5 June 2014