Heat fluxes al the ice-ocean interface and ice thickness are investigated by comparing field data from the Coordinated Eastern Arctic Experiment (CEAREX) drift phase with model calculations. The calculations are based on two types of models. The first one is a one-dimensional ice-ocean model with high vertical resolution. This model is based on the conservation equations for heat, salt, and momentum and uses turbulence models to achieve closure. A discrete element approach is also introduced to explicitly parameterize the ice roughness. The second model is a simple one-dimensional bulk heat transfer model. In this version, the interfacial salinity is modelled on the basis of salt conservation at the ice-ocean interface. The bulk heat transfer model is then calibrated using the former model. The two models predict ocean heat fluxes that are quite variable in time owing to short-term variations in the ice drift. Both models calculate realistic ice thicknesses. It is demonstrated that the observed time variation in ice thickness from eight different experimental sites with varying initial thicknesses and bottom topographies can be reproduced by applying bulk heat transfer coefficients in the range (2.8 +/- 1) x 10(-4). Horizontal variation of the thermal state within a single pack ice floe results in simultaneous freezing and melting over relatively small spatial scales. When modeling or averaging ice data in space these aspects need to be considered.