The measures presented above do not satisfy this condition and thus must be modified. The percentage-measures defining a percentage of still-at-sea were changed to percentage of no-longer-at-sea. The time-measures were changed to one over the measure. To avoid

to large values, a lower limit of 1 hour was used in this study. The measures were not normalized because normalization selleck inhibitor would not affect the routes. However, the actual value of the measure is important for weighting when the measure is not the only term in the target function. The route was found by a very simplistic approach. The route progresses from grid point to grid point of the model grid with a step of at most two. Only eight directions would be possible if only the neighboring grid points were allowed. The number of directions is increased to sixteen by also considering the neighboring grid points one step further. The cost of stepping from one grid point to any of its 16 possible neighboring grid points was calculated in the following way. The direct path between the two grid points passes two or more grid squares. The distance of the

path in each grid square is calculated and multiplied with the value of the measure in the grid square and finally learn more summed up. Dijkstra’s algorithm (Dijkstra, 1959) was used to find a globally optimal path. The mean surface currents for the simulation period are depicted in Fig. 3. In Fig. 4, the investigated measures of this study and the distance to the nearest coast

are shown. The color scale is chosen such that the area of each color is the same in each picture (see figure caption for details) to facilitate comparisons between measures. In the following, higher/lower values refer to higher/lower according to this normalized color scale. The measures form two groups that share many common features. The results of the first group are depicted in Fig. 4a–c, which are the percentage-measures containing information about clonidine the situation after 30 days. The results of the second group are shown in Fig. 4d–e, which are the time-measures containing information about time scales for reaching the 90% level. In the narrow passage between Öland and Gotland, the mean current is directed toward the open area south of Gotland. The percentage-measures are higher than the time-measures. These time-measures are more similar to the distance to the nearest coast. In the more open area northwest of Gotland, the mean current is directed toward Gotland and the south. In this area, the percentage-measures are lower than the time-measures, which in turn are lower than the distance to the nearest coast. In the western Arkona basin, the mean currents are directed toward The west. Here, all measures are lower than the distance to the nearest coast. There are results that divide the measures in different ways, e.g., in the Bornholm Channel, where the mean current is directed toward the Arkona basin.