Best resampling parameters for Biophysical Processor?

Optical Toolbox
1

Hello all,

I’m using the biophysical processor to estimate vegetation indicators over a small area of a L2A product, which I crop using a Subset filter. Since both Subset and Biophysical Processor needs inputs with bands in the same resolution, I’m obliged to use the Resample filter as well.

The question I want to ask here is : does anyone know what would be the best resampling parameters to be used with Biophysical Processor? I would like to have estimates in both 10m and 20m resolutions, so I would be interested in information both up-sampling and down-sampling parameters.

Or maybe it would be interesting to pre-process the product with the recently released Sentinel-2 Super-Resolution?

For information, possible parameters for the resample are (values I’m currently using in bold):

  • Down-sampling:

    • First
    • Min
    • Max
    • Mean
    • Median

  • Down-sampling flag:

    • First
    • FlagAnd
    • FlagOr
    • FlagMedianAnd
    • FlagMedianOr

  • Up-sampling:

    • Nearest
    • Bilinear
    • Bicubic

Best,

1 Like
2

Hello Mellon
I have exactly the same dilemma, several years after the publication of your post. What I can tell you is that I have obtained the best result using ‘Nearest’ because it remains more faithful to the original nature of the pixel value, as opposed to other predictive interpolation methods. What I haven’t been able to find out is which is the best option for “Downsampling methed” and “flag downsampling methed”.