base image mosaic file

[Baldl]

Dear FORCE community,

I encountered an error when doing coregistration, that I don’t really know why or how it happened. I created level 2 Landsat images with the attached params file (landsat_level2) and then a base image with the ls_base file. I do not get an error message and it seems to work fine. However, when I create the mosaic, the file is called:
20130325-20211231_001_365_HL_TSA_LNDLG_NIR_FBM.vrt

Which then leads to the coregistration error: "could not retrieve base image. First 5 digits = 'YYYY-'. coregistration failed."

I noticed that in previous projects I did with FORCE, the naming of the base image mosaic was different (e.g. 2013-2021_001-365_HL_TSA_LNDLG_NIR_FBM.vrt), which would also match the error message from the log files.

I also tried renaming the file according to the old base image I had, but it still gives an error (“coreg (#tie, x/y shift, rmse): 0/0.00/0.00/0.00 coreg failed”). Therefore, I think the root of this error is deeper. When opening the base image mosaic in QGIS, it looks fine with 12 bands, one for each month.

Until now I could not find the error in my processing pipeline, if anyone could give me a hint on where I went wrong with creating the base image, that would be much appreciated.

FORCE v. 3.8.01 on Ubuntu 24.04.2 LTS (GNU/Linux 6.8.0-55-generic x86_64) was used for processing. 32 physiscal cores, 64 threads and 164 GB RAM.

Best,
Lisa

ls_base.txt
landsat_level2.txt

[davidfrantz]

Hi Lisa,

yes, the file naming pattern was changed. Please also see here: #346

Renaming the VRT (but keeping the individual tiff names as they are), should actually work. This was the recommended workaround.

However, as indicated in the linked thread, you can also use the latest dev version, which is able to digest the new naming pattern directly.

Please note that the coregistration can fail if no tie-points are found. This usually happens for a certain fraction of the images and is often linked to cloud coverage.

Best,
David