Version 1 (modified by benj, 10 years ago) (diff)

--

Q: Why are my geocorrected files much bigger than my level 1 files??

A:

Essentially this is a limitation of the BIL format. When lines running NE-SW or SW-NE are mapped and projected, they are diagonal. The BIL format, however, requires rectangular images (and fills in the gaps with zeros), which means your geocorrected images have a high proportion of empty space. Your level 1 files don't have this problem because they're not mapped, and therefore don't require padding to make them square.

There are a couple of possible solutions:

  1. The best solution for working with the images is to process them in pieces (see part 2 here). This means processing the line several times, but the more pieces you use the more you will reduce the total space used (because less blank space will be needed to make the images square - processing in two evenly-sized pieces, for example, will reduce the total size by a half, processing in four will reduce it by 75%).
  1. Because the extra space is all zeros, you should find that compressing the images hugely reduces their size. This will keep them all in one piece, but will probably only be useful if you're archiving them - you'd still have to decompress them to work with them, and you'd spend a lot of time compressing and decompressing.
  1. Converting the images to either HDF 5 or NetCDF 4 format might be a compromise between the two - those include compression (so will save your storage space) and should be able to decompress the data in chunks to allow you to work with it. You'll still lose a certain amount of time to compression and decompression though, and you'll need to find a suitable utility to convert your data.

Back to FAQ


Related articles: