Recent results of satellite observations of sea ice coverage show that sea-ice extent is shrinking by about 3% per decade in the Arctic, whereas it increased by about 1.7% per decade in Antarctica. In the Arctic, observations of decreasing ice extent are accompanied by submarine sonar measurements of ice thickness, showing a thinning of up to 40% between 1958 and today. However, those results are discussed very controversely, because the submarine thickness measurements are restricted in time and space, as they have been performed only in some years, and within a limited region. It is rather likely that due to changes in the atmospheric circulation regimes also known as North Atlantic and Arctic Oscillations (NAO and AO) thicker ice has been moved to other regions which were not profiled by the submarines in later years.

We perform field studies and model experiments to better understand the observed changes and to relate them to possible climate changes.

Figure 1 shows the development of ice volume in the Arctic Ocean between 1951 and 1999, i.e. the product of ice area and thickness. The results have been obtained by means of computations with a numerical dynamic-thermodynamic sea ice model, using six-hourly atmospheric forcing data. The time series is dominated by strong decadal variations which are related to the NAO and AO. The surface winds in response to the sea level pressure distribution cause different ice drift pattern, which determine how much ice is pressed against the coasts of Greenland and Canada to thicken it by deformation, or how long the ice resides within the Arctic Ocean before it is exported through Fram Strait and melts in the North Atlantic. From the figure it is clear that derived thickness trends critically depend on the time period chosen to calculate the trend. Overall, the model yields a reduction of ice volume of 4% per decade.

The limited time and space coverage of submarine thickness data require additional measurements which allow for more systematic thickness surveys. Therefore, we applied and operationalised electromagnetic (EM) induction sounding for ice thickness measurements. This classical geophysical technique is usually employed on land to map ore or groundwater deposits. In the beginning, the accuracy and robustness of the EM method was evaluated by comparing coincident drill-hole and EM measurements. Figure 2 shows two ice thickness profiles thus obtained. To achieve long profiles of some kilometer length, the EM sensor was mounted onto a sledge and man-hauled across ice floes. The figure also shows the good agreement of drill-hole and EM measurements.

Figure 2 includes a representative comparison of two floe profiles obtained in the summers of 1995 and 1996 in the same region of the Laptev Sea. In 1996 the typical ice thickness was 2 m, 50 % thicker than in 1995. It could be shown that the different thicknesses were closely related to different atmospheric circulation regimes in those years, with a strong summer low over the North Pole in 1996, which was completely absent in 1995. The large interannual variability of ice thickness makes interpretation of thickness changes with respect to long-term climate changes involved.