SAR MONITORING OF WAVES IN PANCAKE ICE IN THE MARGINAL ICE ZONE

Abstract

Frazil and pancake ice (FPI) is becoming an important component of the marginal ice zone (MIZ) owing to the dramatic changes occurring in the Arctic seas. This paper deals with the capability of synthetic aperture radars (SAR) to capture the changes suffered by ocean waves traveling in FPI fields, in order to relate them to the thickness of the ice layer crossed. We took advantage of the long-term ERS SAR acquisitions over polar oceans to carry out quantitative SAR spectral analysis for the Odden Ice Tongue (OIT), which developed in the Eastern Greenland Sea in the winter/spring 1997. A sea ice model, specifically developed to predict FPI distribution in the OIT, ran for that period to provide daily ice thicknesses and concentrations over 25 Km spatial scale. In situ samplings and wave ocean spectra collected within FPI fields during an oceanographic campaign for ice-ocean physics study carried on the R/V Jan Mayen were also available. To demonstrate the feasibility of the method, SAR inversion results were compared to the wave spectra collected by a directional wave buoy deployed on different FPI locations. Finally, wave attenuation rates computed over the ERS2 SAR image acquired on March 11, were compared with those collected in the Weddell Sea using an array of wave buoys.

Full text

SAR MONITORING OF WAVES IN PANCAKE ICE IN THE MARGINAL ICE ZONE
Giacomo De Carolis (1), Lucia Maria Laurenza(1)
(1) Istituto per il Rilevamento Elettromagnetico dell’Ambiente, IREA-CNR, Via Alfonso Corti, 12, 20133 Milan, Italy,
Email:decarolis.g@irea.cnr.it
ABSTRACT
Frazil and pancake ice (FPI) is becoming an important
component of the marginal ice zone (MIZ) owing to the
dramatic changes occurring in the Arctic seas. This paper
deals with the capability of synthetic aperture radars
(SAR) to capture the changes suffered by ocean waves
traveling in FPI fields, in order to relate them to the
thickness of the ice layer crossed. We took advantage of
the long-term ERS SAR acquisitions over polar oceans to
carry out quantitative SAR spectral analysis for the
Odden Ice Tongue (OIT), which developed in the Eastern
Greenland Sea in the winter/spring 1997. A sea ice
model, specifically developed to predict FPI distribution
in the OIT, ran for that period to provide daily ice
thicknesses and concentrations over 25 Km spatial scale.
In situ samplings and wave ocean spectra collected
within FPI fields during an oceanographic campaign for
ice-ocean physics study carried on the R/V Jan Mayen
were also available. To demonstrate the feasibility of the
method, SAR inversion results were compared to the
wave spectra collected by a directional wave buoy
deployed on different FPI locations. Finally, wave
attenuation rates computed over the ERS2 SAR image
acquired on March 11, were compared with those
collected in the Weddell Sea using an array of wave
buoys.
1. INTRODUCTION
Frazil and pancake ice (FPI) are two of the most
important sea ice types in the world ocean. FPI are
becoming a relevant component of the Arctic cryosphere,
subsequent to the dramatic decline of sea ice extent and
volume in Arctic seas [1]. The progressive reduction of
sea ice extent allows large waves to develop under the
action of intense winds, which foster FPI development
and thus making them a relevant component of the
marginal ice zone (MIZ). FPI have also an important role
in the Southern Polar Seas. Indeed, it is believed that the
outer portion of sea ice fringing Antarctica continent
during the austral winter consists mainly of consolidated
FPI [2]. Nevertheless, FPI have been very little studied
and their role in the energy system of the global
cryosphere has consequently been neglected. Therefore,
it would be advisable to develop methods for determining
the extent and thickness of FPI. Remote sensing
technology may contribute to achieve this task. The
synthetic aperture radar (SAR) instrument gives the
opportunity to monitor the rapid evolution of the MIZ in
terms of FPI extent and concentration. As SAR proxy for
FPI thickness is not available, a way for its estimation is
given by SAR capability to measure the two-dimensional
ocean wave spectrum that crosses FPI fields. Following
the wave energy decline over the direction of dominant
wave propagation [3,4], such approach may provide the
wave attenuation rates as a function of the frequency
needed to get an estimate of the average FPI thickness.
We demonstrate the feasibility of the method by
discussing results from ERS2 SAR acquisitions over the
Odden Ice Tongue (OIT) in March 1997. OIT is an
occasional sea ice feature, which recursively formed in
the Eastern Greenland Sea during winters 1966–1972,
occasionally built in the 1980s and 1990s and rarely
occurred since 2000 [5]. Fig. 1 shows a map of the area
where the OIT formed in winter 1996/1997. In the
context of this study, the importance of OIT is that FPI
types are the major sea ice components. Then, ERS2
SAR wave attenuation rates were compared to those
measured in the Weddell Sea to estimate the average FPI
thickness [6]. The predictions of a sea ice model which
ran for the winter 1996/1997 over the OIT to provide
daily FPI thickness, concentration and salinity [7] were
used to assess the SAR retrieved FPI thickness.
Figure 1. Odden Ice Tongue region that developed in the
Eastern Greenland Sea in the winter 1996/1997.
2. ERS2 SAR ANALYSIS
Rheological properties (e.g., viscosity, floe size and
distribution) and thickness of an ice layer can alter the
energy spectrum of ocean surface waves. SAR imaging
can capture such changes [3,4]. The non-linear integral
transformation relation proposed by Hasselmann and
Hasselmann (HH91) [8], properly modified to include the
effect of temporal shift between looks [9], was utilized to
_____________________________________
Proc. ‘Living Planet Symposium 2016’, Prague, Czech Republic,
9–13 May 2016 (ESA SP-740, August 2016)

map the two-dimensional ocean wave spectrum into the
SAR image cross-spectrum. Then, the wave energy
spectrum for windows selected at different ranges inside
FPI fields was estimated by inverting the observed SAR
image cross-spectrum according to the SAR inversion
scheme proposed by HH91 [8]. Subsequently, the
relationship between FPI thickness and attenuation rates
as a function of the frequency obtained from wave data
gathered in FPI fields in the Weddell Sea [6], was
employed to estimate the ice cover thickness. Here the
sea ice thickness is defined as the “equivalent solid ice
thickness”:
hf
Vf
ff
hp
Vp
fp
H+=
(1)
where hp (hf) is the thickness of pancakes (frazil), Vp (Vf)
is the volume fraction in pancakes (frazil), and fp (ff) is the
fractional area coverage of pancakes (frazil) [6]. Finally,
results from the SAR retrieval algorithm were compared
with the daily predictions of FPI thickness, concentration
and salinity provided by a salt-flux model specifically
developed for the Odden area, which ran for the
winter/spring 1996/1997 [7].
In this study, the ocean wave spectrum used as first guess
to initialize the SAR inversion procedure was taken from
the positive part of the SAR image cross-spectrum, where
wave signals without directional ambiguity can be clearly
detected owing to the reduced blurring of waves-in-ice
that limits the effects of constructive velocity bunching
[10]. SAR inversion results were assessed by comparing
them with a couple of ocean wave spectra gathered by a
directional wave buoy deployed on March 8 and March
9 1997 on two different FPI locations at the same time of
the SAR passages. The wave buoys were deployed as part
of the field work carried on the R/V Jan Mayen from 3 to
13 March 1997 with the aim of studying the ocean-sea
ice physics in the Odden [11]. SAR inversions were able
to quantitative account for the spectral energy features of
the buoy wave spectra, as will be discussed in the
following sections.
3. 8 MARCH 1997: ORBIT 9840, FRAME 2115
The ERS2 SAR image (orbit 9840, frame 2115) acquired
on 8 March (Fig. 2) corresponds to the first scene that
includes the wave directional buoy. The SAR image
covers both open sea (bright areas) and FPI fields which
appear darker. The buoy was deployed at about 73°N -
6.8°W in an area composed of small pancakes embedded
in dense frazil ice [12]. Fig. 3 reports the map of the
equivalent ice thickness, as computed according to eq.
(1), using the salt-flux model predictions for the same day
of ERS acquisition. Fig. 3 also includes the position of
the selected SAR frame and the ECMWF wind field at
12:00 UTC.
Fig. 4 shows the real and imaginary part of the SAR
image cross-spectrum computed for the sea ice window.
A swell system traveling approximately along azimuth
direction toward North can be recognized.
Figure 2. ERS2 SAR image (orbit 9840, frame 2115)
acquired on 8 March 1997 over Odden area. Buoy
position and selected window for SAR retrievals are
shown along with the ECMWF wind field.
Figure 3. 8 March 1997: equivalent sea ice thickness
[cm] from the salt-flux model. Red polygon represents
SAR frame. The ECMWF wind field is also shown.
Fig. 5 and Tab. 1 summarize the main SAR inversion
results for the selected window. The ERS2 SAR image
measured a significant wave height (Hs) of 5.82 m, to be
compared with Hs = 5.43 m from wave buoy. The
resulted correlation between observed and simulated
SAR spectra was as high as 0.98.
4. 9 MARCH 1997: ORBIT 9854, FRAMES 2114 -
5890 - 2141
In order to better identify the ice edge, three consecutive
ERS2 SAR images were considered (orbit 9854, frames
2114, 5890, 2141), the wave buoy being imaged within

the frame 5890 (Fig. 6). The darker area along the ice
edge can be assigned to frazil formation while in the
northernmost frame different texture and brightness
signatures, presumably related to different mixtures of
FPI can be distinguished.
Figure 4. 8 March 1997: real (left) and imaginary (right)
part of image cross-spectrum computed for sea ice
window#1.
Figure 5. 8 March 1997: SAR inversion results for
window#1. Blue line of the bottom right panel represents
the buoy wave spectrum.
Table 1. Inversion results from SAR image of 8 March
1997, win#1
Several windows were selected within open sea and sea
ice regions. The buoy was deployed at about 73°N - 1°W
and window#23 was approximately centred around it.
Fig. 7 shows the sea ice map of equivalent thickness
along with the position of the selected frames. Fig. 8
shows the real and imaginary part of the SAR image
cross-spectrum for sea ice window#23.
Figure 6. Sequence of ERS2 SAR images (orbit 9854,
frame 2114, 5890, 2141) acquired on 9 March 1997.
Buoy position and window#23 centred around it are
shown along with the ECMWF wind field.
Figure 7. 9 March 1997: equivalent sea ice thickness
[cm] from the salt-flux model. Red rectangle includes
SAR frames of Fig. 6. The ECMWF wind field is
superimposed.
Fig. 9 and Tab. 2 summarize the main SAR inversion
results for window#23. A wave height Hs = 4.29 m was
estimated from SAR, which is very close to Hs = 4.24 m
recorded by the wave buoy. Also in this case we obtained
a correlation value between observed and simulated SAR
as high as 0.99.

Figure 8. 9 March 1997: real (left) and imaginary (right)
part of image cross-spectrum computed for sea ice
window#23.
Figure 9. 9 March 1997: SAR inversion results for
window#23. Blue line of the bottom right panel
represents the buoy wave spectrum.
Table 2. Inversion results from SAR image of 9 March
1997, window#23
5. 11 MARCH 1997: ORBIT 9883, FRAME 2115
The ERS2 SAR image (orbit 9883, frame 2115) acquired
on 11 March (Fig. 10) includes a large field of FPI
showing the typical swirling pattern caused by the local
forcing of wind and waves on the small cakes. Frazil ice
is also visible as dark feature separating open sea (bright
area) from the remaining FPI fields. For this SAR
acquisition co-located wave buoy data were not available
as the nearest buoy gathered data about 3° SW from the
southernmost corner of the SAR image. Two different
windows were selected inside the ice field, far from the
ice edge, namely window#1 and window#10, shown in
Fig. 10. The corresponding map of the salt-flux model
equivalent ice thickness is shown in Fig. 11.
Figure 10. ERS2 SAR image (orbit 9883, frame 2115)
acquired on 11 March 1997 with the position of the
windows selected for ice thickness estimation. The
ECMWF wind field is also shown.
Figure 11. 11 March 1997: equivalent sea ice thickness
[cm] from the salt-flux model. Red polygon and arrows
represent SAR frame and ECMWF wind field,
respectively.
SAR retrieved wave spectra are consistent with a wind
wave field traveling from the open sea as the retrieved
wave height is reducing going from window#10 to
window#1. Figs. 12 and 13 summarize the relevant SAR
inversion results. These allowed computation of the
energy wave attenuation rates in order to compare them
with those obtained from an array of wave buoys in the
Weddell Sea [6]. In this case the sea ice thickness was
about 8 cm, which is close to the average value of 11 cm
predicted by the sea ice model for the FPI strip between
the couple of selected windows. Fig. 14 shows the result
of this comparison.

Figure 12. 11 March 1997: SAR inversion results for
window#1.
Figure 13. 11 March 1997: SAR inversion results for
window#10.
Figure 14. Comparison of SAR inverted attenuation rates
with those obtained in the Weddel Sea [6], as a function
of the wave period.
6. SUMMARY
In this study we exploited ERS2 SAR image spectra
observed in the OIT on 8, 9 and 11 March 1997 to map
the change in the wave spectrum in terms of energy
attenuation according to the SAR inversion scheme
proposed in [8]. The sea ice SAR inversion procedure
was assessed by comparing co-located wave spectra
gathered by a directional wave buoy deployed on March
8 and March 9, 1997, respectively. SAR wave retrievals
compared favourably with wave buoy spectra as well as
observed and simulated SAR image spectra show very
good agreement, with correlation values of 98% and
99%, respectively. Finally, for the case of 11 March, the
SAR inverted attenuation rates are consistent to the
equivalent sea ice thicknesses predicted by the sea ice
model developed for the OIT. The latter results should be
regarded as preliminary as ongoing research in this
context is being carried out in order to assess the
capability of SAR imaging to retrieve FPI ice thickness
from SAR spectral image analysis.
Acknowledgments We thank Leif Toudal Pedersen
(DTU) for having provided the salt-flux model results.
The work was carried out within PANACEA project
(MIUR, grant 2013/AN2.02) and the FP7 EU project
“Ice, Climate, Economics - Arctic Research on Change”
ICE-ARC (grant agreement 603887) http://www.ice-
arc.eu/. The ERS2 SAR images were provided by ESA
within the project ID 31587 “Comparison of SAR frazil
and pancake sea ice thicknesses with predictions from a
salt-flux model in the Marginal Ice Zone”.
7. REFERENCES
1. Feltham D. (2015). Arctic sea ice reduction: the
evidence, models and impacts, Philosophical
Transactions. Series A, Mathematical, Physical, and
Engineering Sciences, 373(2045), 20140171.
http://doi.org/10. 1098/rsta.2014.0171
2. Wadhams P. (2000). Ice in the ocean. Australia:
Gordon and Breach Science Publishers
3. Raney R. K., P. W. Vachon, R. A. De Abreu and A.
S. Bhogal (1989). Airborne SAR observations of
ocean surface waves penetrating floating ice, IEEE
Trans. Geosci. Remote Sensing, vol. 27, no. 5, pp.
492-500
4. Wadhams P., F. Parmiggiani, G. De Carolis, D.
Desiderio and M. J. Doble (2004). SAR imaging of
wave dispersion in Antarctic pancake ice and its use
in measuring ice thickness, Geophys. Res. Lett., 31,
L15305, doi: 10.1029/ 2004GL020340
5. Rogers, J. C. and M.-P. Hung (2008). The Odden ice
feature of the Greenland Sea and its association with
atmospheric pressure, wind, and surface flux
variability from reanalyses, Geophys. Res. Lett., 35,
L08504, doi: 10.1029/2007GL032938
6. Doble M. J., G. De Carolis, M. H. Meylan, J.-R.
Bidlot and P. Wadhams (2015). Relating wave
attenuation to pancake ice thickness, using field

measurements and model results, Geophys. Res.
Lett., 42, doi: 10.1002/ 2015GL063628
7. Toudal Pedersen L. and M. D. Coon (2004). A sea ice
model for the marginal ice zone with an application
to the Greenland Sea, J. Geophys. Res., 109, C03008,
doi: 10.1029/2003JC001827
8. Hasselmann K. and S. Hasselmann (1991). On the
nonlinear mapping of an ocean wave spectrum into a
synthetic aperture radar image spectrum and its
inversion, J. Geophys. Res., vol. 96, no. C6, pp.
10713-10729
9. Engen G. and H. Johnsen (1995). SAR-ocean wave
inversion using image cross spectra, IEEE Trans.
Geosci. Remote Sensing, vol. 33, pp. 1047–1056,
1995
10. Lyzenga D. R., R. A. Shuchman and J. D. Lyden
(1985). SAR imaging of waves in water and ice:
evidence for velocity bunching, J. Geophys. Res., vol.
90, no. C1, pp. 1031-1036
11. Wadhams P., J. P. Wilkinson, R. J. Hall and S. J.
Down (1997). Ice-ocean physics cruise to the Odden
region of the Greenland Sea, Report of the ESOP-2
project of the EU MAST-III programme
12. Toudal Pedersen L., P. Wadhams, J. Wilkinson, M.
Doble, H. S. Andersen, F. Parmiggiano, G. de
Carolis, T. B. Loyning, A. Rudberg, T. Vinje, J.
Kolatschek, T. Martin, M. Kreyscher, T. Jakobsson,
J. E. Wallevik and B. Uscinski (1999). Sea ice
physics, in: The thermohaline circulation in the
Greenland Sea – Final scientific report, The
European Sub Polar Ocean Programme, phase 2,
MAST III contract MAS3-CT95-0015