Mapping the Thickness of Pancake Ice Using Océan Wave Dispersion...
25
as another method of estimating h. However, in most cases a frazil-pancake icefield has a very ragged or imperfectly defined edge; it is seldom straight.
Therefore the angles of incidence and refraction of a wave cannot be defined
accurately, so this method is usually inapplicable.
4 Results from Arctic Areas
4.1 Chukchi Sea
The first dataset to be analysed in this way was a Seasat image from the Chukchi
Sea northwest of Wrangel Island, obtained in October [12]. Fringing the main
Arctic Océan ice edge was pancake ice, with frazil ice seaward of it. Six sets of subscenes were analysed, along lines running parallel to the main wave vector into
the ice. Wavelengths involved were in the range 150-180 m. Spectra were unimodal
and clearly defined. Refraction in the correct sense was observed in every case.
The wavelength changes were used to compute ice thickness, which increased linearly to 2-3 m at 25 km inside the ice edge and 4-6 m at 50 km inside the ice edge.
These values were felt to be high, but not necessarily impossibly high, since clearly this type of ice had amassed against the thicker polar ice due to the action of a
strong on-ice wind, which might hâve piled it up in a wedge which could hâve
reached such a thickness. The absence of any in-situ vérification was a drawback
here. Only the simple spectral analysis was carried out, without any attempt at
inversion.
4.2 Greenland Sea
During the 1993-1996 ESOP (European Subpolar Océan Programme) Project, an
EU MAST-2 project to study the rôle of frazil-pancake ice in stimulating
Greenland Sea convection, extensive direct observations were carried out in the
Odden ice tongue, a protrusion of ice, usually frazil-pancake, which grows eastward from the main East Greenland ice edge in winter in the vicinity of latitudes
72-75°N [18]. These measurements included ice thickness measurements, both of
the pancakes and of the frazil ice matrix. Most measurements were made on the
well-developed 1993 season Odden, which provided an opportunity to make comparisons with the wave analysis of concurrent ERS-1 SAR images. The work was
done in two stages. Initially, ail possible ERS-1 imagery of the 1993 Odden was
acquired and examined. In every case where the image showed a clear wave train
entering an edge of Odden, the image was subjected to a straightforward spectral
analysis. The nature of the changing SAR spectrum as the waves entered the ice
was studied in order to assess the conditions under which the simple technique
may be considered to hâve some validity. Finally, the full inversion technique was
applied to selected data in order to examine the changes which this produces in
the spectra.
As an example, the inversion algorithm was applied to an ERS-1 SAR PRI
image of the Odden ice tongue in the Greenland Sea acquired on April 10, 1993
25
as another method of estimating h. However, in most cases a frazil-pancake icefield has a very ragged or imperfectly defined edge; it is seldom straight.
Therefore the angles of incidence and refraction of a wave cannot be defined
accurately, so this method is usually inapplicable.
4 Results from Arctic Areas
4.1 Chukchi Sea
The first dataset to be analysed in this way was a Seasat image from the Chukchi
Sea northwest of Wrangel Island, obtained in October [12]. Fringing the main
Arctic Océan ice edge was pancake ice, with frazil ice seaward of it. Six sets of subscenes were analysed, along lines running parallel to the main wave vector into
the ice. Wavelengths involved were in the range 150-180 m. Spectra were unimodal
and clearly defined. Refraction in the correct sense was observed in every case.
The wavelength changes were used to compute ice thickness, which increased linearly to 2-3 m at 25 km inside the ice edge and 4-6 m at 50 km inside the ice edge.
These values were felt to be high, but not necessarily impossibly high, since clearly this type of ice had amassed against the thicker polar ice due to the action of a
strong on-ice wind, which might hâve piled it up in a wedge which could hâve
reached such a thickness. The absence of any in-situ vérification was a drawback
here. Only the simple spectral analysis was carried out, without any attempt at
inversion.
4.2 Greenland Sea
During the 1993-1996 ESOP (European Subpolar Océan Programme) Project, an
EU MAST-2 project to study the rôle of frazil-pancake ice in stimulating
Greenland Sea convection, extensive direct observations were carried out in the
Odden ice tongue, a protrusion of ice, usually frazil-pancake, which grows eastward from the main East Greenland ice edge in winter in the vicinity of latitudes
72-75°N [18]. These measurements included ice thickness measurements, both of
the pancakes and of the frazil ice matrix. Most measurements were made on the
well-developed 1993 season Odden, which provided an opportunity to make comparisons with the wave analysis of concurrent ERS-1 SAR images. The work was
done in two stages. Initially, ail possible ERS-1 imagery of the 1993 Odden was
acquired and examined. In every case where the image showed a clear wave train
entering an edge of Odden, the image was subjected to a straightforward spectral
analysis. The nature of the changing SAR spectrum as the waves entered the ice
was studied in order to assess the conditions under which the simple technique
may be considered to hâve some validity. Finally, the full inversion technique was
applied to selected data in order to examine the changes which this produces in
the spectra.
As an example, the inversion algorithm was applied to an ERS-1 SAR PRI
image of the Odden ice tongue in the Greenland Sea acquired on April 10, 1993