adeane-ga
commented
3 years ago Just placing this here for additional context This is the email chain from Urs Wegmuller (GAMMA) regarding this issue.
Dear Alistair,
If I understand right you determined the Insar phase in the -PI,PI interval and then geocoded this. This phase has of course jumps and so the interpolation is problematic.
I strongly recommend that you geocode the complex valued interferogram and then you calculate the phase after that. So the interpolation is a well posed problem and you get an excellent solution. You can use the default method.
Kind regards,
Urs
Am 01.11.2021 um 23:55 schrieb Alistair Deane:
> Hi Urs,
>
> Thanks for getting back to me.
>
> You are right, and I have found the source of our issue, but do have some follow up questions. To see the range, I am querying the files manually rather than using a display command.
>
> The source of our issue is in the "geocode_back" step when taking the real valued interferogram and georeferencing it. See attached image for some print out of data range produced from intermediate steps for the flattened interferogram example. The respective GAMMA command is commented above the results.
>
> In our current workflow, we have default interpolation settings (bicubic spline):
>
> geocode_back 20201007-20201031_VV_8rlks_flat_int_flt 8630
> 20201019_VV_8rlks_geo_to_rdc.lt geocode_flat_ifg.tmp 12737
>
> When I change to Nearest Neighbour, we see suitable values again in the wrapped interferogram:
>
> geocode_back 20201007-20201031_VV_8rlks_flat_int_flt 8630
> 20201019_VV_8rlks_geo_to_rdc.lt geocode_flat_ifg.tmp 12737 - 0 0 - - -
> 0
>
>
> So this is good news, except I wonder what your thoughts are on this?
>
> Specifically:
> - Are we using this correctly, would you recommend a different interpolation method or sequence of commands?
> - I do notice that Nearest Neighbour is not recommended according to the GAMMA documentation and tests.
> - Perhaps a higher order Bicubic Spline (or other method) could work, but I am hesitant if it can produce values outside of the +/- pi range (I haven't yet tested a higher order or the other methods).
>
> Also, this our sequence of GAMMA commands relating to this part of the interferogram processing for the Flattened example:
> 1. SLC_diff_intf
> 2. cpx_to_real
> 3. geocode_back
> 4. mask_data
>
> Let me know if I can collate more information.
>
> Thanks,
> Alistair
>
> -----Original Message-----
> From: Urs Wegmuller <wegmuller@gamma-rs.ch>
> Sent: Friday, 29 October 2021 6:28 PM
> To: Alistair Deane <Alistair.Deane@ga.gov.au>
> Subject: Re: SLC_diff_intf [SEC=UNOFFICIAL]
>
> Dear Alistair,
> the resulting wrapped interferogram is containing complex values, so the phases are by definition only between +/- pi.
>
> How do you display the data to get phase values between -4.3 and +4.3
> radians (what command / program are you using ?)
>
> Kind regards,
> Urs
>
> Am 29.10.2021 um 04:39 schrieb Alistair Deane:
>> Hi,
>>
>> My team here are using GAMMA and we have noticed that our wrapped
>> interferograms are containing data values that are greater than +/- pi.
>> So we are getting a range approximately between -4.3 and +4.3 radians.
>>
>> We are most likely using a parameter setting incorrectly or using the
>> wrong sequence of GAMMA programs. Could you advise us on whether you
>> know what might be causing this?
>>
>> This is how we are forming the initial flattened interferogram:
>>
>> SLC_diff_intf r20201007_VV.slc r20201031_VV.slc r20201007_VV.slc.par
>> r20201031_VV.slc.par 20201007-20201031_VV_8rlks_off.par
>> 20201007-20201031_VV_8rlks_sim0_unw
>> 20201007-20201031_VV_8rlks_flat0_int
>> 8 2 1 0 0.25 1 1
>>
>> Thanks,
>>
>> Alistair
>>
>> *Alistair Deane *|InSAR Scientist
>>
>> National Geodesy Section | Place, Space and Community Division
>>
>> t: +61 2 6249 9818 www.ga.gov.au <http://www.ga.gov.au/>__
>>
>>
>> Geoscience Australia Disclaimer: This e-mail (and files transmitted
>> with
>> it) is intended only for the person or entity to which it is addressed.
>> If you are not the intended recipient, then you have received this
>> e-mail by mistake and any use, dissemination, forwarding, printing or
>> copying of this e-mail and its file attachments is prohibited. The
>> security of emails transmitted cannot be guaranteed; by forwarding or
>> replying to this email, you acknowledge and accept these risks.
>
--
**************************************************
Urs Wegmuller
GAMMA REMOTE SENSING Email: wegmuller@gamma-rs.ch Worbstrasse 225, CH-3073 Gümligen, SWITZERLAND
Tel: +41(0)31-951.70.05 Fax: +41(0)31-951.70.08
http://www.gamma-rs.ch
*************************************************
mwheeler
Summary
geocode_backstep, based on the testing (shown below).Details The issue is that the
geocode_backstep contains interpolation. We are using the default interpolation method Bicubic Spline, and so the values are not constrained to +/- pi during the interpolation.Solution - to be tested After emailing GAMMA, they recommend that we do the
geocode_backstep prior to the interferograms being converted to real values - so prior to thecpx_to_realstep.Here is what they say:
Testing ---Initial Flattened IFG (complex), SLC_diff_intf--- Working on: 20201007-20201019_VV_8rlks_flat_int
MIN: -3.1415927410125732 MAX: 3.1415927410125732 MEDIAN: 0.0
---Initial Flattened IFG (real), cpx_to_real--- Working on: 20201007-20201019_VV_8rlks_flat_int_flt
MIN: -3.141592502593994 MAX: 3.141592502593994 MEDIAN: 0.0
---Initial Flattened IFG (real, geocoded), geocode_back--- Working on: geocode_flat_ifg.tmp
MIN: -4.565230846405029 MAX: 4.448446750640869 MEDIAN: 0.0
---Initial Flattened IFG (real, geocoded, masked), mask_data--- Working on: 20201007-20201019_VV_8rlks_flat_geo_int
MIN: -4.565230846405029 MAX: 4.448446750640869 MEDIAN: 0.0