Land on subset area of oil spill

hai, i am a newbie here to process data with sentinel 1 and snap.
i am interest to detect oil spill using sentinel 1 with tools oil spill detection on snap, the area is in a bay, so land still found there.
is existence of land influence process and result of oil spill detection ?

thank you

Did you see that several tutorials on this topic exist?

i did see…
i mean, even if the image had land masked, is algorithm will produce different result when the area at a bay and offshore ?

sorry, don’t understand the question.

Sorry for my messy English :pray:
I think what i mean is, where is “alook alikes” more often appears? Onshore or Offshore ? Thank you

You mean if you can search for oil spills over land areas? The SNAP module only works over water bodies but is quite reliable as long as the water is calm and the surface is smooth.

Yes, on the beach maybe… so the module just for water bodies? What is the formula behind algorithm?

yes, open water only.

You find some information in the Help section:

The paper mentioned there might be more explanative:
https://ieeexplore.ieee.org/abstract/document/1370264

I did read help windows, it give an logic explain, i mean formula like in wind field estimation. Hehe
Thank you for your journal recomendation Mr.ABraun, i will read to understand more :pray:

I’m afraid there is no formula which directly translates SAR backscatter into classified pixels of oil. It is rather based on moving window and thresholding techniques.

okay Mr, thank you for the answer.
i want to ask another question in this topic. what is the reason to put “background window size” value ? how to determine appropriate background windows size ?

some considerations on the correct utilization and selection of a background window size are given here: Oil Spill Graph - Detection/Cluster Output Looks Wrong

Oke, thank you for your recomendation

Sorry for another topic i’va made, i think its different topic :pray:
So is it allowed to use same value of background windows size and Threshold on multitemporal data ?

I’d say your results are most comparable if the same parameters are used for every image. Also radiometric calibration is advisable for any multi-temporal analysis.

Sorry lack of focus :pray:
Thanks a bunch