As OC algorithms sometimes use spectral bands (MODIS for instance) with different spatial resolution, I would like to know which technical solution is used to merge different spatial resolution.
If spatial resolution is degraded (from finer to coarser), how is the default aggregated value computed ? This would affect the final result of OC algorithm result - max value of Rrs for instance or in case of border effect.
If spatial resolution is improved (from coarser to finer), which is the default interpolation method used ?
The question of spatial interpolation can also be asked regarding atmospheric correction of remotely sensed signal (MODIS again) using very coarser meteorological reanalysis data.
- User Services
- Posts: 1442
- Joined: Wed Sep 18, 2019 6:15 pm America/New_York
- Been thanked: 1 time
While it is true that we produce and distribute remote sensing reflectance data for MODIS that include the higher spatial resolution bands (469nm, 555, and 645nm), none of the ocean color products we produce use bands other than the nominal 1km spatial resolution ("ocean") bands.
For the higher resolution Rrs bands reported at 1km, the bands are aggregated to match the nominal 1km footprint (it is really only ~1km near nadir). The l2gen code does have the ability to generate 250m and 500m outputs for MODIS, and when it does, the lower resolution bands are replicated (oversampled) though this is not done for the products we distribute.
As for the handling of the lower resolution ancillary inputs needed for the atmospheric correction process, those fields are bilinearly interpolated to the senor resolution. For some ancillary fields a linear temporal interpolation is also done.