For pansharpened imagery, what preprocessing method to do first- is it atmospheric correction or orthorectification? From my readings, it seems the former but the rational polynomial coefficients or RPCs needed for orthorectification are then removed in the corrected imagery which makes it impossible to do the latter. If the reverse is done with orthorectification first, wavelengths needed for atmospheric correction are removed as well making the latter impossible to accomplish. What gives?

  • 2
    Please edit your question and provide more information. In particular, what software are you using, what satellite/sensor and format is your imagery?
    – user2856
    Jun 18, 2015 at 4:03

1 Answer 1


In most processing chains, orthorectification or reprojection is performed as late as possible in the workflow, to avoid altering the spatial relationships between neighbouring pixels.

In your case, re-sampling first would alter the value of pixels say, at the interface between a water body and land, which would then lead to incorrect estimates in the algorithm used for atmospheric correction.

It sounds like it's more a technical issue of passing the required metadata to be able to perform the RPC ortho on the atmospheric corrected image. If you are using the GDAL Python bindings, it is possible to copy the metadata domain from the source image and set it in the corrected output dataset using something like :

ds.SetMetadataItem("ERR_BIAS", myoriginalbiaserror, "RPC")

for each required metadata item. See https://trac.osgeo.org/gdal/wiki/rfc22_rpc for the full list of RPC metadata items.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.