What is residual height?
May 9, 2015 at 5:19 am #853
first of all I should appreciate your great software. It is my pleasure to work with such a software.
I like to know more about height and residual height.
what should we do if the residual height is high in some areas? is anything wrong with the parameters which I have taken or the TDX DEM which I have used? Is it anyway to reduce these residuals? are these residuals after removing look angle error?
May 9, 2015 at 5:39 am #854
I have another question. what should I do if the coherences are not good after APS removal? what are DSF and R parameters in APS options box?
May 9, 2015 at 6:51 am #855
Residual Height = Height – External DEM
Removing an external DEM should reduce the unknown height (with “unknown height” I mean the source of the topographic component of the interferometric phase). Removing an external DEM is particularly useful in mountainous areas (the external DEM can decrease significantly the range of the unknown height). It may be useless in urban areas. In fact, most of the geometric component of the phase in a urban area is usually generated by the height of buildings, and usually an external DEM is not detailed enough to give you the building height.
Moreover, even if you would have a very precise urban 3D model, it’s basically impossible to retrieve the height of scattering centers from it (a scattering center can be formed e.g. by a balcony, or a fence, or by the orientation of tiny metallic details…).
If you have high values of residual height this may be due to many reasons:
– you are processing many uncoherent points
– the range of res. height to estimate you have choosen is too big
– your DEM is misaligned
– you really have a lot of high/low structures
It’s difficult to judge without looking at your data: you could post here the results of your estimation.
About your second question, check this: http://www.sarproz.com/groups/sarproz-group/forum/topic/aps-estimation/
if the temporal coherence of the processed connections is good but after estimating and removing the APS it is not good anymore, it means that the APS is badly estimated. Probably then your non-linear weights are causing a bad graph inversion. Again, it would be good if you could post here the histograms and also the colored graph (where the color shows you the connections coherence).
May 10, 2015 at 1:26 am #857
here are some files of the process.
after APS removal the coherence of each interferogram became almost zero except 2 of them.
I think I have done something wrong during the process. but I don’t know where is my mistake.
- This reply was modified 2 years, 1 month ago by farnoshhosseini.
Attachments:You must be logged in to view attached files.
May 10, 2015 at 1:52 am #865
can you include a picture of the APS processing module, so we see which options you picked up?
tell us also which images graph you are using for the estimate
May 11, 2015 at 3:36 am #870
I used default processing parameters. but in non linear weighting I used 0.6,0.8,1 accordingly. I select inverse parameter method to remove APS. the chosen network was MST one sensor.
now I am switching to small area mode to check the parameters range
May 11, 2015 at 4:13 am #871
which factors should we check to the ranges which we have selected for the parameters? how can we make sure that which range is proper?
May 11, 2015 at 8:11 am #872
first very important comment
why are you using the MST?
I’d suggest you to work with the Star Graph.
In fact, you are analyzing a structure, not distributed targets.
I also suggest you not to use weights.
Try this, and then let me know.
As for the range of parameters is concerned, at the beginning you can ignore it and use the default values.
Then, based on the result, you can judge whether you have to modify it or not.
Elements which can help you in the choice are:
– size and shape of the peak of the histogram of the estimated parameter
– external infos
– coherence values.
It would be very helpful to see some results of the small area processing over your bridge (using star graph, no weights)
You must be logged in to reply to this topic.