Some anomalies occured when using Polymer4.13 on OLCI

Post Reply
Knap
Posts: 1
Joined: Mon Apr 12, 2021 4:54 pm
company / institution: HoHai University
Location: NanJing

Some anomalies occured when using Polymer4.13 on OLCI

Post by Knap »

Hello fsteinmetz!

Some blocky and striped anomalies appear when using Polymer4.13 on OLCI. These anomalies seem to usually occur in turbid waters. I displayed the results in the figure and marked them with a black box. I would like to ask what caused this phenomenon. Can I avoid it by adjusting some parameters? At present, my parameters are all using the default.
poly.png
In addition, I also used C2RCC to process the same image, and found that the difference between the two algorthms is quite large, both in clean (like Sea of Japan, Ploymer seem to be the larger one) and turbid(off-shore, Polymer is alwaye the smaller one) waters. I made sure that both displays use the same scale and both display water leaving reflectances at 510nm.
c2rcc.png
Regards,

tian
You do not have the required permissions to view the files attached to this post.
User avatar
fsteinmetz
Site Admin
Posts: 306
Joined: Fri Sep 07, 2018 1:34 pm
company / institution: Hygeos
Location: Lille, France
Contact:

Re: Some anomalies occured when using Polymer4.13 on OLCI

Post by fsteinmetz »

Dear Tian,

Many thanks for your feedback.
Indeed this is a known issue, caused by occasional unstable processing over turbid waters. These bad pixels should be flagged (see bitmask). Inverstigations are ongoing to improve these results, but unfortunately this is not going to be solved in the short term.
The vertical striped are due to the fact that the pixels are processed in the vertical direction first, and the initialization of each pixels is based on the result of the previous pixel, leading to vertical patterns in case of unstability. These anomalous patterns will look different (ie, not vertical anymore) if you activate the option force_initialization=1 in run_atm_corr. However, processing will be a bit longer in this case, and anomalous pixels will remain.

Also, there are indeed quite some differences between C2RCC and Polymer. It is also known that Polymer underestimates the reflectance of extremely turbid waters, so C2RCC may be better in this case. However, Polymer results over clear waters should be quite good ; the dark Pattern on C2RCC in the sea of Japan seems a bit suspect and could be due to sun glint or aerosol plume.

Kind regards,
François
Post Reply