CSKS georeferencing error

s1tbx
#1

Hello everyone,

I’m processing a CSKS [Processing Level: SCS] image with SNAP (8.0.2) (Desktop). I performed the following operators:

  • Multilook
  • Calibration
  • TC
    However, the processed image is not correctly georeferenced, being shifted by 13 km north from the real position. I don’t think the problem is TerrainCorrection, indeed, also just opening the raw product on SNAP the world view window shows me the footprint in the wrong position (more north than real):

image
image1033×684 120 KB

It is not the first time I have this kind of issue with CSKS images in the last weeks (months).

Does anyone have an idea of the source of the problem?

Many thanks

#2

Just to be clear, some of these products are alright and others are not?

1 Like
#3

Dear @ABraun, exactly, I have had many of these problems in the last months with commercial SAR image processing in SNAP. This particular product is a very high resolution image (around 0.7 m), but I don’t think this is a relevant factor.

I also realized that in the metadata of the product in question are reported the 4 geographic coordinates of the bounding box of the scene acquired (through the tags GeoCoordinTopRight/Left…BottonRight/Left) and these seem to coincide exactly with the incorrect positioning of the processed image (I mean that the provider has acquired the correct area of interest but has been given slightly wrong geographic coordinates … this slightly translates, however, in 13 km of shift).

Therefore, in this case, I believe that the problem could be with the RAW product and not with SNAP.

Do you think this could be a common situation with commercial products?
Do you think that the only solution to the problem is a subsequent coregistration of the image?

#4

If the bounding box is already falsely stored in the metadata, I would try to get in contact with the data provider.

You can try the SAR Simulation coregistration, which is more suitable when geocoding is not perfect, but it needs a DEM of comparable quality

1 Like
#5

Dear Mr. @ABraun, thanks for the advice, I think that’s the best thing.

As for the SAR Simulation coregistration, unfortunately I can only dream of a DEM with a resolution of about one meter over the area of interest :sweat_smile:

#6

true that…
Maybe if you try the coregistration of the products as initial step and select Orbit as Initial Offset Method. If we are lucky, you can use the correctly geocoded product as a reference and the others as support products and the orbit information will be more correct than the bounding box in the metadata.

1 Like
#7

Thank you very much Mr. @ABraun for your usual kind support. I will try to apply your suggestion. However, my primary goal was to be able to solve the problem without the need to use the correctly geodcoded image (which in this case I only have for luck).