32
P. Wadhams et al.
analysis was applied to the other two SAR scenes of April 24 (orbit 5289, frames
5571 and 5589).
The very long wavelength observed in the spectrum is typical of a swell wave
System for which the methodology to dérivé ice thickness developed in the preceding sections cannot be applied. The high wind speeds and effectively infinité
fetches typical of the Southern Océan generate much wave energy at long periods; at the same time, as we showed in Section 1.2, pack ice acts as a low pass filter, allowing only the longest waves to pass through it. Thus an open water environment separated from the main Southern Océan by a wide belt of pack ice, as
occurred at the time of these images, is a geometry which guarantees that only
long period waves will be présent, for which the dispersion relation is not sensitive to ice thickness. This technique is not applicable in such circumstances. We
do, however, consider that it may be applicable to the outer ice edge.
6 Discussion
Inspection of cases where "impossibly large" changes in wave number occur on a
simple analysis show that what is actually happening is that an azimuthal component is being suppressed in the open sea and is then appearing within the ice,
a conséquence of the azimuthal cut-off effect combined with lower energies
inside the ice. This indicates the need to carry out the full inversion in such cases;
the simple method can only be considered valid when the main wave direction
both inside and outside the ice is close to the range direction.
The best controlled and clearest cases of changed wave dispersion in the
Greenland Sea still give excessively high values for ice thickness. This agréés with
the high results found by Wadhams and Holt [12]. It is clear from Greenland Sea
results that the changed wave dispersion is indeed in the right direction (towards
larger wave numbers), but the simple mass loading theory predicts ice thicknesses which are greater than those really occurring. Two possibilities présent themselves. One is that the problem is just due to the failure to carry out a full inversion, i.e. that nonlinear effects may be moving the position of the main wave peak
such as to exaggerate the real wave number change that is occurring. The other is
that the theory is inadéquate and requires modification. It is possible, for instance, that the sédimentation equilibrium of the frazil ice suspension in the
water column gives it a greater effective dynamic thickness than the thickness
that is measured when the frazil is scooped from the water surface. In situ underwater video observations of frazil ice in a wave field would be a way of determining this.
Winter conditions in the Greenland Sea or Southern Océan may be
unfavourable much of the time for the optimal use of this method. The reason is
that wave energies are high (which is good for imaging the waves), but at the
same time fetches are large and wind speeds are high and thus dominant wave
periods are high. At high wave periods the change in wave number on entering
an icefield is low, according to the mass loading theory. Ideally we need shorter
waves - although not too short, because of the minimal resolution due to pixel
size. For instance, in ail 33 cases analysed in the Greenland Sea, the dominant
P. Wadhams et al.
analysis was applied to the other two SAR scenes of April 24 (orbit 5289, frames
5571 and 5589).
The very long wavelength observed in the spectrum is typical of a swell wave
System for which the methodology to dérivé ice thickness developed in the preceding sections cannot be applied. The high wind speeds and effectively infinité
fetches typical of the Southern Océan generate much wave energy at long periods; at the same time, as we showed in Section 1.2, pack ice acts as a low pass filter, allowing only the longest waves to pass through it. Thus an open water environment separated from the main Southern Océan by a wide belt of pack ice, as
occurred at the time of these images, is a geometry which guarantees that only
long period waves will be présent, for which the dispersion relation is not sensitive to ice thickness. This technique is not applicable in such circumstances. We
do, however, consider that it may be applicable to the outer ice edge.
6 Discussion
Inspection of cases where "impossibly large" changes in wave number occur on a
simple analysis show that what is actually happening is that an azimuthal component is being suppressed in the open sea and is then appearing within the ice,
a conséquence of the azimuthal cut-off effect combined with lower energies
inside the ice. This indicates the need to carry out the full inversion in such cases;
the simple method can only be considered valid when the main wave direction
both inside and outside the ice is close to the range direction.
The best controlled and clearest cases of changed wave dispersion in the
Greenland Sea still give excessively high values for ice thickness. This agréés with
the high results found by Wadhams and Holt [12]. It is clear from Greenland Sea
results that the changed wave dispersion is indeed in the right direction (towards
larger wave numbers), but the simple mass loading theory predicts ice thicknesses which are greater than those really occurring. Two possibilities présent themselves. One is that the problem is just due to the failure to carry out a full inversion, i.e. that nonlinear effects may be moving the position of the main wave peak
such as to exaggerate the real wave number change that is occurring. The other is
that the theory is inadéquate and requires modification. It is possible, for instance, that the sédimentation equilibrium of the frazil ice suspension in the
water column gives it a greater effective dynamic thickness than the thickness
that is measured when the frazil is scooped from the water surface. In situ underwater video observations of frazil ice in a wave field would be a way of determining this.
Winter conditions in the Greenland Sea or Southern Océan may be
unfavourable much of the time for the optimal use of this method. The reason is
that wave energies are high (which is good for imaging the waves), but at the
same time fetches are large and wind speeds are high and thus dominant wave
periods are high. At high wave periods the change in wave number on entering
an icefield is low, according to the mass loading theory. Ideally we need shorter
waves - although not too short, because of the minimal resolution due to pixel
size. For instance, in ail 33 cases analysed in the Greenland Sea, the dominant