The first screenshot below is of the oceancolor website for this granule. The chlorophyll image looks right (i.e., similar in geographic extent to that on the maps on the right side, as well as the extent shown on the initial granule selector /search page). The RGB image, however, looks to show a different extent.
If I download the L1B and process to L2 (all default parameters in L2GEN), the output appears to be only part of the full granule (only 641 lines appearing to be the northernmost ones). The second screenshot is the chlor_a field in the SeaDAS GUI, with just the LAND l2flag applied (the Outer Banks are clearly recognizable to the right). If I squint, I can see that the RGB image on the ocean color site is just this region, stretched vertically.
So, it looks like both my local processing and that which created the RGB image on the ocean color site produce the same result, while whatever created the chlor_a image on the ocean color site has a different extent. Thoughts?