shift between TanDEM-X and external DEM

Added by Sofia Antonova about 6 years ago

Hi,

I have several TanDEM-X CoSSC pairs and I'm trying to obtain differential interferograms using airborne laser scanning DTM.
These are my steps:

gc_map (to calculate lookup table and DEM related products)
pixel_area (to generate simulated SAR image in range-doppler coordinates rather than geocode)
gc_map_fine (refinement)
pixel_area (pixel area correction factor calculated using the refined lookup table)
geocode (resample the DEM into radar range-doppler coordinates)
phase_sim_orb (#simulate the interferometric phase)
SLC_diff_intf (#calculate the differential interferogram)

Results are slightly different for each pair. For some pairs there is a residual shift between TanDEM-X and ALS DTM, and for some pairs the shift is not visible. I tried to improve the refinement of the lookuptable many times - not successfully. Next idea is to coregister TanDEM-X pairs to each other, but I'm not sure that this is a right solution? That would mean to coregister SLCs to each other, meaning to lose the initial coregistration of CoSSC pairs.

Thanks a lot for any help!
Sofia

Replies (3)

shift between TanDEM-X and external DEM - Added by Charles Werner about 6 years ago

Hello,

Did you make sure that the external DEM has ellipsoidal heights (NOT GEOIDAL)?
The reference ellipsoid for the GAMMA programs such as phase_sim_orb is WGS84.

My understanding is that the TanDEM-X data pairs are already coregistered.

You should be able to refine the lookup table for each pair by measuring the
offsets between the simulated scene and the actual. It is possible to use
exactly the
same DEM area for the DEM segment by specifying the segment DEM_par and Segment
data, then gc_map will not try to come up with another bounding box.

You would terrain geocode the first (master) pair, and then all other pairs
would be geocoded to exactly the same area.
The refinement is based on using an offset measured using offset_pwr between the
scene and the simulated scene, and then applying the change to the lookup table
using gc_map_fine.

You can perform this process using the script mk_geo_radcal. There is HTML
documentation for mk_geo_radcal viewable with the command
gamma_doc mk_geo_radcal

Best regards,
Charles

RE: shift between TanDEM-X and external DEM - Added by Sofia Antonova about 6 years ago

Dear Charles,

thanks a lot for your reply.

The heights of external DEM are above WGS84 ellipsoid, I made sure.

TanDEM-X pairs are not coregistered. There is a significant mismatch between some pairs. Therefore, I produce and refine the lookuptable for each pair separately.
The problem starts at the generation of differential interferogram - some pairs still show a visible shift w.r.t. the external DEM. Because some other pairs do not show a shift I thought I might coregister all TanDEM-X pairs to the one which doesn't feature the shift. But the command SLC_diff_intf produces the differential interferogram directly from SLCs, meaning that I would need to coregister SLCs, disturbing the precise coregistration of CoSSC product.
Alternatively, I could use SLC_intf, then coregister the obtained interferogram to the "good" one, and then use sub_phase to subtract the external DEM. Or this way is not good for TanDEM-X bistatic mode?

Thanks!
Sofia

RE: shift between TanDEM-X and external DEM - Added by Charles Werner about 6 years ago

Hello,
Each scene can be generated in its own geometry, but then terrain geocoded to
the same DEM segment. The different height maps from different pairs are only
combined when they are geocoded.
That is what I am proposing.

Take the reference scene of each pair (master) and terrain geocode it, also
generating the initial DEM in radar geometry (mk_geo_radcal -d option).
Make sure to use the same DEM segment for all pairs. No SLC resampling required.

best regards,
Charles

(1-3/3)

Project

General

Profile

GAMMA Questions