Abstract. Small-, medium-, and large-scale measurements of sea ice surface roughness were carried out in the Baltic Sea over a 3-year period. A clear relationship was found between the rms height and correlation length of Baltic sea ice. In most cases the surface correlation function was close to exponential. The logarithm of rms height showed a linear dependence on that of measured distance. Similarly, the correlation length was in many cases linearly dependent on the measured distance. This dependence on length suggests that ice surface roughness could be characterized as a fractal. Surface roughness is usually described with rms height •r and correlation length L. Actually, these are pseudoparameters, as they depend on the measured length [Church, 1988]. For conventional surfaces the dependence is sometimes so weak that these parameters can be taken as constants, but this is not so with fractal-like surfaces. Rms height and correlation length can be determined also for fractal-like surfaces but will not then represent intrinsic surface properties and will depend on the measurement process used. Fractal surfaces will typically have measured correlation lengths, which are of the same order as the measured length no matter what length is used. In nature the fractal characteristics of a surface will not extend to infinity; there is an upper limit to the measurement length, the surpassing of which leads to saturation of the rms height and correlation length values. The surface then behaves like a random Gaussian surface with a very high maximum surface height difference. The adequate surface roughness description to be chosen depends on the application in question. If the surface area to be characterized is smaller than required for the saturation of the conventional roughness parameters, the effect of measurement length cannot be neglected.
Field ExperimentsThe years 1992 and 1993 were very mild, making ice measurements possible only in the Bay of Bothnia (Figure 1). In 1994 the ice extent was mostly very close to average for the Baltic Sea, and measurements were accordingly carried out in the Bay of Bothnia, the Sea of Bothnia, and the Gulf of Finland. The ice types were also much more varied in 1994 than during the previous 2 years.In 1992, measurements were carried out mostly in ridged old ice areas, with ice blocks in the ridges some 40 cm thick. Because the whole ice field was deformed repeatedly, no large level ice areas were available for comparison. In 1993 the study area was the edge of an old, many times deformed ice field (Figure 2). Here the ice block thickness was typically about 20 cm.In 1994, surface roughness measurements were carried out on old level ice and deformed areas in the Bay of Bothnia and the Gulf of Finland. The ice field of the Sea of Bothnia was newer and had deformed radically about 1 week prior to the field experiment. The deformation appeared in the Bay of Bothnia and the Gulf of Finland either as long ridge sails, rubble fields (Figure 3), or consolidated pancake ice. The ice in ...