Parallel_stereo with mapprojected images over glaciers

17 views
Skip to first unread message

Arnab Sarkar

unread,
May 13, 2026, 1:11:53 PM (13 days ago) May 13
to Ames Stereo Pipeline Support
Hi, I had been using normal stereo and parallel_stereo, but as I used parallel_stereo with mapprojected images, I found that the steep slopes get captured better with fewer voids in the DEM. 

I was wondering about an issue, as I am using Copernicus or SRTM as a reference DEM (by first converting its datum to ellipsoid). My DEM contains hilly slopes and glaciers. So, when I am mapprojecting my current images with a DEM which is old enough (of 2000 or 2015), will my glacier elevation get biased by the reference dem elevation at the glacier? or will it not be impacted by the reference? 

As a glacier's elevation changes with time, so does using mapprojected images with the reference DEM of a different time a correct approach?

Oleg Alexandrov

unread,
May 13, 2026, 3:15:43 PM (13 days ago) May 13
to Arnab Sarkar, Ames Stereo Pipeline Support
>will my glacier elevation get biased by the reference dem elevation at the glacier? or will it not be impacted by the reference? 

That is quite possible, especially if your glacier does not have much texture. Then stereo correlation, especially with the asp_mgm algorithm, could perhaps find some textures that comes from the underlying mapprojection DEM, which would be incorrect.

It is suggested to blur the mapprojection DEM quite a lot, and also try this experiment with different DEMs or at least different level of blur. 

--
You received this message because you are subscribed to the Google Groups "Ames Stereo Pipeline Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ames-stereo-pipeline...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ames-stereo-pipeline-support/509a12a8-9790-431f-9878-ac0875e3c8b6n%40googlegroups.com.

Arnab Sarkar

unread,
May 13, 2026, 3:34:13 PM (13 days ago) May 13
to Ames Stereo Pipeline Support
Thanks for your quick response.

I will definitely try with a more blurred and multiple levels of blurring. I had a connecting query to this. Do the elevation of the changed surface for my case, the glacier be shifted to the reference DEM one? I mean a vertical shift or bias, or will it be calculated based on the given new fore and aft mapprojected tifs and RPC files?

Appreciate your thoughts on this.

Regards,
Arnab

Oleg Alexandrov

unread,
May 13, 2026, 3:48:33 PM (13 days ago) May 13
to Arnab Sarkar, Ames Stereo Pipeline Support
>Do the elevation of the changed surface for my case, the glacier be shifted to the reference DEM one? I mean a vertical shift or bias, or will it be calculated based on the given new fore and aft mapprojected tifs and RPC files?

Ideally both of these should be true. So yes, your glacier elevation should not depend on reference DEM, and it should be calculated based on new mapprojected tif files.

So mapprojection DEM is just a helper. But in practice, as before, this one can bleed over into the solution, which is undersriable. 


Reply all
Reply to author
Forward
0 new messages