Eagle/Hawk spectrums misaligned

[benj]

If you plot Eagle and Hawk processed output on the same graph (see attachment) there's about a 30nm overlap for the spectral range, and the instruments don't seem to agree in this overlap. Even if the instruments don't give the same magnitude, the absorption features shown should align (and they don't). This suggests that the spectral calibration of one or both instruments may be out by up to 100nm, which appears to be the displacement needed to make the two graphs align.

[benj]

Eagle/Hawk spectra from level 1 data (Nigg Bay line 3). Top lines are over land, bottom over water (with possible sun glint).

[benj]

Eagle and Hawk spectra over land and water processed by Caligeo

[benj]

Data processed by Caligeo shows a similar pattern. However, I'm starting to think that the spectral calibration isn't actually off very far but instead the Hawk radiometric calibration is wrong. If you look at the overlapped area, on the land spectra for both az* and Caligeo there is a (smallish) absorption feature on the Eagle spectrum right above one in the Hawk spectrum. The assumption had been that the Hawk feature matched up to the Eagle one to the left and that the spectral calibration was wrong. However, it actually makes more sense if it matches the one above (ie the spectral calibration is correct) and the magnitude is significantly wrong (ie the radiometric calibration is wrong).

If the hawk feature matched the eagle feature to the left then you would expect another absorption feature in the Hawk spectrum further up, and there isn't one. In addition, the wavelength of the large absorption feature in the Hawk spectrum matches that of the small, high one in the Eagle spectrum quite closely.

[benj]

Output of SMARTS and Hawk (8px av. over land)

[benj]

Output of SMARTS and Eagle (8px average over water)

[benj]

Eagle and Hawk spectra both appear to have correct spectral calibration. Of two graphs below, both can be seen to show correctly-positioned absorption features - Eagle is pretty much exactly on, Hawk is ~5-6nm high but since the comparison data is modelled rather than actual this is not unreasonable.

Note for both graphs, magnitude of sensor output is different to modelled output - this is expected, scale is different

Over water (easiest to see Eagle):

Over land (easiest to see Hawk)

[benj]

Closing this ticket since spectral calibration seems fine, issue with potential iffy Hawk radiometric calibration continued in ticket #113