Within March 2013 a prototype of an ice route optimization system was tested within a two weeks trial in Barents Sea using research ship Lance. The system is based on two main components, an ice forecast and a route optimization finding the fastest routes through varying ice conditions depending on the ice manoeuvring capability of a specific ship. After giving a short introduction to the ice forecast and navigation model the paper will mainly adress the findings on the systems capability to predict the ship during the trial based on a comparison to observations and measured data.

On the trial optimized route options were provided for ten different route proposals in the region east of Svalbard between 76° N and 79° N. The routes were specified by starting and endpoint. The route optimization used an ice forecast from a regional model with a spatial resolution of 5km to minimise the traveling time starting from the shortest distance between start and destination point. The main benefit of the model is given by the inclusion of temporal change of ice conditions driven by wind and currents. The optimization algorithm determines the attainable speed on a route option by time iteration including the actual time step of forecast output between two waypoints. Each waypoint may then be moved along gradients of ice thickness and ice concentrations such that the overall distance of the route will increase but the travel time will be decreased due to higher attainable speeds in lighter ice conditions.

In order to assess the ice conditions on the trial permanent observations of the ice coverage as well as ice and snow thickness in the viscinity of the ship were carried out and documented. Additionally the ice thickness was measured using an electro magnetic induction device (EM31). At the same time the ship navigational data like speed, course over ground and percentage of engine power were documented.

During the trial the system proofed to be a useful assistance for navigation enhancing the radius of information on the ice situation beyond the range of radar and visibility. Different routes proposed by the system could be tried out while in several cases the ship first tried to follow the shortest distance option and was then forced to switch to one of the neiboughring route options in lighter ice conditions.

As the optimisation is computed on a grid with 5x5km2 cells, deviations between predicted and obtained travel time on the routes in ice can be related to the effect of averaging the ice conditions and speed over one grid cell. As the data observed and measured during the trial are given at a sampling rate of two minutes they can be used to quantify this effect by providing fine scaled distributions of ice parameters and ship speed for each chosen track.

For further development of the system the shortcomings given by the limited number of ice parameters included in the ice forecast could be determined and ideas to improve the quality of speed prediction by using implicit information about the distribution of certain ice parameters within one grid cell may be proposed.

This content is only available via PDF.
You can access this article if you purchase or spend a download.