Algorithm for validation of distribution of first year ice
Validation vs. ice type from satellite data
The observational product includes an ice type defined as Relatively young ice, which has here been taken to represent first year ice. The ice type variable in the observations have been gridded so that each grid has been assigned with one ice type only (i.e., the observational product does not allow for a mix of different ice types in the same grid cell).
The model product includes gridded values for fraction of total ice cover in each grid cell (fSI), and values for fraction of first year ice in each cell (fFYI). The grids cells in ice infested waters where
fSI - fFYI < 0.5
are taken to represent the area which is infested with first year ice.
The two products used here are available with slightly different horizontal resolutions. Thus, prior to the analysis, the model product is projected onto the observational grid.
Note that during summer, extending into the season with minimum sea ice extent in the Arctic, the algorithm to produce ice type data from the satellite observations yields a dominance of cells with an ambigous ice type or low levels of confidence. Consequently, the areal extent for which the validation is performed has a strong seasonal cycle.