Valuation of ice compression hazard by means of fuzzy logic model

Ice compression is serious navigational hazard for all ships proceeding in ice. In most cases no obvious indications of compression precense can be observed from the ice surface, the phenomenon could activate and develop very fast being sometimes local both in time and space. The classical methods of ice dynamics observations and the modeling describing the compression with one physical parameter, the internal stress are not able to evaluate the degree of threat for ice compression in particular sea area (in space or time or both). Therefore, alternatively it is probably meaningful to describe the ice compression with an integrated parameter (not directly measureable), which is divided into classes according to the degree of risk. We suggest to realize this approach using fuzzy logic model based on parameters obtained from numerical atmospheric and ice dynamic models (typically meso-scale) and/or different other sources of information about the forcing, also non-physical and empirical knowledge and data, observed in ship scale. First, we take into account parameters which make up the potential (or power) of compression. Among these parameters are ice thickness, ice compactness and ice categories (e.g. ridges etc). Second, we consider parameters which trigger the potential to the compression hazard. For the simplest model such triggering parameters could be wind forcing pattern and its change or some other estimates obtained from the relevant model output and/or remote sensing products. A fuzzy logic model considers both the potential and triggering measures and provides a forecast of hazard of ice compression at a given grid point (or at a certain location) for particular individual ship with its navigational parameters. The model output is appropriately divided into classes like missing, mild, moderate or severe compression. For the model calibration and validation the data from ocean circulation models which include the ice sub-model (e.g. HIROMB) or pure ice dyn- mics models (e.g. HELMI) could be used along with the relevant data from ship traffic systems (AIS), direct observations at ship bridges, classical ice maps, satellite images and ice drifter data.