Earthdata Forum

Hello All,
I see some papers by NASA researchers suggesting 3X3, 5X5, 7X7 window size when validating atmospherically corrected images with in situ Rrs. If using validation data from e.g., AERONET-OC, and I have a lots of masked pixels in my 7 X 7 window, can I go above that - say, using  11 X 11 or more?

The size of the box should be sufficiently large to have statistical confidence in the retrieval, but small enough to ensure that the average measurement is representative of that seen by the in situ data.  If the waters are homogenous, you can use a larger region.  If there is a lot of local variability, using a larger region may degrade the validation results.  AERONET-OC sites tend to be coastal, and as such may have higher local variability.  Keep this in mind when interpreting the results.

Sean

Thanks so much Sean!