thank you, @heinzollerketchup.
Generally, the longer the temporal baseline (=the time between two images) the higher is the decorrelation which leads to small coherence values. It also depends on the cover of the analyzed surface. Bare ground mostly shows slower decorrelation than one with high vegetation cover.
thanks again, your post just explains my second question. My study area (in Mongolia) and has a low coherence even in a 12 day time gap (mean 0.4). So according to your response this result is influenced by the surface (shrubs, woodland, grasland, bare soil) and not by the time interval.
Is there a possibility to increase the coherence in areas with covered earth surface?
The wavelength of Sentinel-1 is about 5 cm and strongly interacts with smaller vegetation cover such as shrubs. Ich these shrubs slightly change between two image acquisitions coherence rapidly decreases.
It cannot be increased by methods of data processing. Higher coherence can be achieved through precise orbit parameters, shorter time intervals and longer wavelengths (ALOS PALSAR with L-band, ca. 21 cm is less sensitive to smaller surface cover changes).
So all you can do with sentinel is applying the orbit files before your co-registration (also increase the number of GCPs) and hope that the coherence is of better quality.
"Select a point where you assume no change and read the value. Let’s say its 45 cm
Subtract 45 from your imagein order to have zero values at those areas with no change. The remaining variation should be due to subsidence between two overpasses."
can you explain it briefly … you say “45 cm”… from where I get this value for my case… is it from phase or coherence …from coherence map> properties> pixel information tab>band… little confused
ok I understood … but ABraun can you show me procedure of subtraction step wise … that you mentioned
"Subtract 45 from your imagein order to have zero values at those areas with no change. The remaining variation should be due to subsidence between two overpasses:
if you expect no subsidence at this location it means that your subsidence map is 208 mm too high. If there was no change at this pixel there should be a zero. After unwrapping your product a product of relative changes within the raster, given as absolute values (mm in this case if you applied the formula correctly).
So make your raster 208 smaller and you should theoretically have the absolute height changes during the two overpasses.
However, there’s a strong ramp in your image wich distorts the result massively. I don’t think you did anything wrong so far but maybe there’s too much time between your images.
i am trying to unwrapped the phase on VMware … it seems a never ending process… it shows treesize 33876567 , pivots 1376897546, improvements 11567898… what does these term means… is there any way to know how much time it takes…
you said that the crop possess of eliminating the pixel that have low coherence will be done on unwrapped phase … but when i do this process on unwrapped phase … a error message is shown…
but when i did it on coherence image only the required pixel remain no error message shown… is it correct…
I can’t help you when I don’t know the error message.
As you work with the information of the unwrapped phase you need to mask out the locations with low coherence there, of course. It’s kind of map algebra with two rasters: If coherence is low, set unwrapped phase to nan/invalid.
Wavelength is 3.1 cm - I didn’t find it on the rush in the metadata, sorry.
incidence angle is dependent from the distance to the sensor.
So there is always a difference between the far and the near range.:
You find it under Abstracted_Metadata:
