I calculated the surface area for a small water body (first, I extracted the water body from the image) using VV and VH polarizations of the Sentinel-1A mission in time series. The results showed that VV polarization is more accurate than VH polarization. In some months, VH polarization resulted in a relatively larger difference with in-situ surface areas, but VV polarization behavior was more stable.
In your opinion, Why do you think VH polarization has more difference than VV polarization in some months with in-situ surface areas?
So VV leads to better quality in all cases? How was the water body extracted (classification, thresholding,…).
As VH polarization is generally lower (except for some vegetation bodies), I expect more contrasts in VV polarization and therefore better separability. Another reason could be the impact of inundated vegetation at the edges of the water body (which one by the way?)
VV is also reported to be less affected by incidence angle (here)
thank you dear Andreas for comments,
so i using the thresholding method for water body extracted with median filter.
And, VV in some month better quality than VH (For example, in the first months of the year, such as January, February, May and April).
it could as well result from a slightly rougher surface (caused by wind), independently from the season. Also, the signal-to-noise ratio (SNR) is often higher in VV than in VH.
How large is the water body?
thank your for your good comments dear andreas, they helped me a lot.
so, the area of water body from 24 km^2 to 43 km^2 is changeable in different years.
I see. So it is possible that the accuracy differences could also have a random component?
Dear andreas yes, sometimes the difference is positive, sometimes negative. but with VV it’s 90% negative with in-situ areas.
that means the thresholding under-estimated the water area?
Maybe you can use the mask manager based on both VV and VH to combine criteria of both polarizations.
Some explanations are given here: Synergetic use of S1 (SAR) and S2 (optical) data and use of analysis tools (page 23).