Issue on POLYMER OLCI

[Fluquorr]

Hi all,

There is an issue occurred when I processed the polymer-corrected OLCI files. Here are some examples of Chla and SPM plots on Jan 14th and 20th, 2024. I found that there are some low-value pixels occurred with a plume along the southern shore of Long island (See attachments). In the Rrs figures, you can also see the same issue. Lastly, I obtained the Rrs spectra curve which is a pixel in that low value area, you can see that the feature is consistent with Rrs figures, Seems POLYMER is overestimating the blues, and underestimating the reds. This issue do not show on other atmospheric correction methods.
The band is being flagged by POLYMER for various reasons, including THICK_AEROSOL, OUT_OF_BOUND, HIGH_AIR_MASS and ANOMALY_RWMOD_BLUE (the bitmask values in the plume outskirts are all >5000). For this example I showed, I have checked and the pixels is flags = 6144 = 4096+2048 = INCONSISTENCY+ANOMALY_EWMOD_BLUE. The figures keeps pixels that are included in the valid pixel expression as defined by the POLYMER developer. So there is a way to mask them out, but those should be valid pixels. So I think this issue is worth posting a question.
Looking forward to your reply.

Screenshot 2024-04-15 at 4.10.36 PM.png

download.png

image.png

Best,
Tong

[fsteinmetz]

Hello,

Apologies for the late answer and thanks for your feedback.
This kind of algorithm unstabilities happen undersome circumstances, here most likely the presence of turbid waters and the grazing sun angles.
The flags INCONSISTENCY+ANOMALY_RWMOD_BLUE are not recommended for systematical use for OLCI, but here they seem useful for detecting the bad pixels, so the flagging strategy may need to be reconsidered.
Also, you may want to try the experimental "minabs2" configuration introduced in Polymer v4.15, which can be activated with some configuration (see v4.15 changelog) - this mode should improve the results in these conditions.
Kind regards,
François