In my area (Northwest Atlantic) PAR is missing in areas of seasonal ice. The algorithm can't differentiate between bright pixels caused by cloud and bright pixels caused by surface conditions, so it relies on ancillary data for ice cover.
Since you mention monthly data, I assume you are working with level-3 files. You may find it instructive to gather all the level-2 files for an area with missing data so you can examine level-2 flag values (e.g., using the SeaDAS 8 GUI) for a few pixels that are missing data.
SeaWiFS coverage was shifted seasonally away from the winter pole, so if you are using SeaWiFS data there won't be any data for polar winter areas.