I am analyzing a number of PACE AOP granules, Level 2 v3, e.g. PACE_OCI.20240708T202028.L2.OC_AOP.V3_0.nc
and am seeing significantly negative Rrs values at ~690nm (see Attachments).
I expect it is a mis-correction of the O2 B-Band in the atmospheric corrections.
If so, is there a mitigation?
Especially troublesome is that the Rrs uncertainty is very small at these wavelengths, i.e. the negative values are >5 sigma departures from 0. This implies the data are highly confident to have measured negative Rrs (non-physical).
Thanks!
X
PACE OCI Feature at ~690nm
Re: PACE OCI Feature at ~690nm
The negative Rrs values observed near 690 nm are primarily due to imperfect correction of the Oxygen B-band absorption feature. Mitigation efforts are underway to better address these residual artifacts, including the use of a more recent absorbing-gas database (HITRAN 2024 instead of 2020), as updates to spectroscopic line parameters are expected to improve the accuracy of the atmospheric absorption correction.
Regarding uncertainty, the current V3.1 processing uses pre-launch top-of-atmosphere uncertainty values, which primarily represent instrument radiometric uncertainties and do not fully account for uncertainties associated with absorbing-gas correction. In a subsequent version of the AOP products, the uncertainty framework will be updated to incorporate post-launch, on-orbit calibration information as well as model and algorithm uncertainties, following the approaches described in:
https://doi.org/10.1364/OE.460735
https://doi.org/10.1364/OE.502561
https://doi.org/10.3389/frsen.2025.1670390
In the meantime, caution should be exercised when using Rrs near 690 nm in very clear waters, where the signal is inherently small and close to zero, making it particularly sensitive to residual atmospheric-correction errors.
Regarding uncertainty, the current V3.1 processing uses pre-launch top-of-atmosphere uncertainty values, which primarily represent instrument radiometric uncertainties and do not fully account for uncertainties associated with absorbing-gas correction. In a subsequent version of the AOP products, the uncertainty framework will be updated to incorporate post-launch, on-orbit calibration information as well as model and algorithm uncertainties, following the approaches described in:
https://doi.org/10.1364/OE.460735
https://doi.org/10.1364/OE.502561
https://doi.org/10.3389/frsen.2025.1670390
In the meantime, caution should be exercised when using Rrs near 690 nm in very clear waters, where the signal is inherently small and close to zero, making it particularly sensitive to residual atmospheric-correction errors.