This is in continuation of my Q/A with the OC Forum about using gpt to mosaic MODIS and OLCI images. In the end I started using IDL because while gpt provides mean reflectances we are interested in composting 7-day images to find mean, minimum and median reflectances for a region.
Since I want to exclude pixels with certain L2 flags, I use L2 data we generated from l2gen. I composite (spatial mean) reflectance first over a regular grid and then find time mean, minimum, and median. My question is about reprojection of L2 data. generally I would use gpt to reproject data in geographic projection. But now I simply bin data in regular lat-lon grids -- as I need to keep track of flags before getting the statistics. Is that OK?