ABSTRACT The topography along the ice water interface can be recovered from airborne electromagnetic data With an Iterative inversion scheme. This method is based on a new fast algorithm for solving the forward problem, which can be used to evaluate electromagnetic scattering by an irregular but highly conductive surface. INTRODUCTION The accurate measurement of floating ice thickness has been a matter of considerable interest to the scientific community for many years. In general this geophysical parameter is of great importance for global climatic studies. On a more limited scale, it is a factor in the safe and efficient operation of icebreakers and other shipping. Finally, on a local scale, knowledge of the sea ice thickness and its distribution is of vital importance m the planning and installation of oil drilling platforms. Because of the wide spread interest in this problem,. the possible means for effecting a remote measurement of sea Ice thickness has been the subject of periodic reviews. All of these, including a recent one prepared by Canpolar Associates (Canpolar, 1985) for the Canadian Department of Fisheries and Oceans have indicated that the required measurements could be made either with low frequency radar or with a conventional airborne electromagnetic exploration system. Of the two, the latter is possibly preferable, because radar is susceptible to erroneous reflections from same moisture zones within sea ice. It appears that a confidential proposal to useairborne electromagnetic (AEM) for Ice thickness determinations was lodged with the US Navy as early as 1968 (Kovacs et al., 1987). That proposal and other studies that followed (e.g. Becker et at, 1983) clearly indicated that no difficulties were to be anticipated in obtaining the average thickness of sea Ice m areas where the ice-water interface was relatively smooth. These predictions were clearly borne out by recent experimental work. Here we must include a demonstration by Won and Smiths determine its height above sea water. Taking test data from Cape Cod, they demonstrated that this can be done with an accuracy of about 30 cm. Allowing for an observational error of about 10 cm for the laser altimeter which is used to map the elevation of the upper snow covered ice surface, we find that one can expect the difference of these two measurements, which yields the ice thickness, to be accurate within about 40 cm. In this context, Kovacs et al. (1987) reported the results of a test on Prudhoe Bay, Alaska which was done over an ice floe where the ice thickness varied from about two to six meters over a traverse distance of about 250 m. Here, the true average ice and snow thickness of 3.62m was recovered from the airborne data with an accuracy of about 50 cm or roughly 15%. It is worthwhile to note that the largest error in the ice thickness estimation was made over an ice keel which had an average thickness of about 5m.