I am trying to coregister an ERS and an S1 image, both SLC with precise orbits applied and with the S1 image deburst. However, I always end up with a completely blank slave image.
Settings as follows:
Initial offset - tried both orbit and product geolocation
Resampling - tried nearest neighbour and bicubic interp.
Everything else is defaults.
Thanks for your help!
you can try the Cross InSAR Resampling module and use ERS as a master and Sentinel-1 as a slave. I haven’t tried but it works with ERS/ENVISAT coregistration.
Tried it, unfortunately no joy. It only lets me convert ERS to the Envisat file format, did that and attempted the coregistration once again, with the same result. Is there a way of converting/resampling ERS/Envisat to the S1 format?
did you try DEM assisted coregistration?
Let me see…different orbit track, different pixel size & resolution, probably clearly different incidence angle (depends on orbit track), and slightly different center frequency -> these images have very different geometry and cannot be co-registered.
What are you trying to achieve? Since the geometries are incompatible interferometry is impossible. If you just want to co-locate the imagery generate intensity-images, terrain correct them separately with a DEM…then you might be able to co-register them into a stack.
Thanks! DEM assisted coregistration seems to work, however I get a small offset which isn’t ideal, but I will keep tinkering around the edges and see if I can sort it out.
Marcus, I’m not worried about the pixel to pixel relationship. It’s change detection that I’m after, purely based on backscatter intensity. Call it geo-co-location if you like. Hence as long as I can remove the geometric distortions, resolution, pixel phase, orbit tracks and incidence angles are pretty much irrelevant for this purpose.
Also, the correction using a DEM shouldn’t really be necessary, as the area I am looking at is as flat as a pancake. However, there might of course be processing reasons that make the use of a DEM necessary…
in this case I would also suggest Terrain Correcting first. Afterwards you can stack the bands or still coregister. Doesn’t matter if you have no topography, Range Doppler Terrain Correction projects your data into a coordinate reference system which can then be used for your purpose.
Cheers! Will try and report back…
This seems to do the trick, many thanks Andreas and Marcus for the help. It is greatly appreciated.
I have the same problem.But my data is ERS/Envisat.Here is my data:
All my dataset cover the period from 1992 to 2003. The amount of data is too large for the computer to process all at once. So I registered in batches.The result of coregistration has a blank slave image --Intensity_slv3_26July1992,intensity of others slave image can be displayed normally.I’m following this tutorial:http://step.esa.int/docs/tutorials/S1TBX%20Stripmap%20Interferometry%20with%20ERS%20Tutorial.pdf
I have attached my picture, Could you please give me some advice? Thanks !
were orbit files applied to the input data?
An empty slave band means not enough matching points (GCPs) were found for the pair.
ERS can be tricky to coregister, please try the DEM assisted coregistration.
I’m sorry not to have given you more details.Orbit files were applied to the input data and GCPs of this batch is 2000(default), it generated an error like this "The specified region, if not null, must intersect with the image`s bounds"when I set 3000. The other batches is ok,I don’t konw the DEM assisted coregistration,but I will try it.