I am processing nine RS2 images to study temporal deformation over an area. In APS module > Sparse point selection group, I have selected Amp. Stb. Index > 0.6 and plotted PSC. The figure shows some horizontal and vertical lines along which the PS density is high. I could not see any kind of such pattern in my reflectivity map and I am wondering what those horizontal and vertical lines are?
If the software is processing entire image by dividing it into several sub-images, is there a chance that more PS points are being identified where two corresponding patches join together?
I am attaching snapshots of reflectivity map & PSC plot. Could you help me interpret what those lines are?
yes, the lines appear because the image is subdivided into patches. however, they look like lines in the picture that you attached, but in fact it’s just an optical effect showing the higher density close to patch borders. if you zoom in, you will see that they still pick up real local maxima, and pixels are not aligned, so, you do not need to worry about that.
I would worry on the contrary about what you are doing: you are processing 9 images and you are using indexes which are not stable at all with only 9 images (e.g. amplitude stability index > 0.6 is actually mostly noise, and this is why you see the patches effect in your selected points). with 9 images you may need to apply some spatial filtering (and maybe even unwrapping) on a special set of interferograms to reach the proper stability.
How should I apply spatial filtering to interferograms? What options I should choose?
Usually I follow the below procedure:
1. Generate interferograms (flat earth phase and DEM removal, goldstein phase filtering, no unwrapping)
2. APS processing (one-star graph, amplitude stability index, estimate linear trend and height, using external DEM, no weights, inv. residuals)
3. Sparse point processing (similar options as above)
As per your suggestions, unwrapping interferograms at step1 would help me in identifying proper stable pixels. Am I correct?
As the no. of images are less, will my results improve if I choose non-star graph, higher value of amp.stb.index and by using coherence as weights?
1. if you process the STAR graph and you do not use the coherence as weight, there is no need to generate interferograms (the sw uses unfiltered phase)
2. if you have few images, a possibility to make the system more stable is processing a non-star graph to add redundant interferograms. if you do so, you may want to filter interferograms and use the coherence as a weight
3. unwrapping is an additional option. if you can unwrap well your interferograms then you can deal with unwrapped phase and you do not need to worry anymore about the stability of the system. however, this works only if the unwrapping is good.
4. choices/options are up to you in connection with your site, your area, your dataset. you need some experience to make the best choices. but it’s a good training.
read this: http://www.sarproz.com/wp-content/papercite-data/pdf/perissin-rsiwpct2012.pdf
As per your suggestion, I have processed RS2 images (HH polarization) and got some results. I have observed subsidence over urban areas. Now I want to process same images with VV polarization to see how different would be the results. But when I tried this, slaves are not being extracted. The software shows ‘preparing extracted slaves for viewing’ and at last it shows ‘slaves not extracted’. Could you please help me.