1

Topic: spot 6

Hi dear Francois
I have two Spot 6 level 1 pan data and I've followed any steps of Cosi-Corr manual.
but I have some problems still:
1- The DEM I used for process is Aster 30m but the original images (spot 6) has 2m resolution.
By using these data, the output of resampling step is completely black, why do you think?
but when I cliped DEM to a smaller area (by subset), it worked and I could continue. is this true?
2- After above steps I went to correlation with different window size but I couldn't get a good result, is this because of low resolution of DEM?
3- My working area is mountain and I saw that in GCP optimization and also resampling step, Bilinear method is better than sinc method, have you ever had any experience with it?
best regards
anna

2

Re: spot 6

Hi dear Francois
I have two Spot 6 level 1 pan data and I've followed any steps of Cosi-Corr manual.
but I have some problems still:
1- The DEM I used for process is Aster 30m but the original images (spot 6) has 2m resolution.
By using these data, the output of resampling step is completely black, why do you think?
but when I cliped DEM to a smaller area (by subset), it worked and I could continue. is this true?
2- After above steps I went to correlation with different window size but I couldn't get a good result, is this because of low resolution of DEM?
3- My working area is mountain and I saw that in GCP optimization and also resampling step, Bilinear method is better than sinc method, have you ever had any experience with it?
best regards
anna

3

Re: spot 6

Hi Anna,

if you are in a mountainous area, depending on the incidence angles of your two images, you'll definitively going to have topographic residual in your coregistration. The 30m GDEM is too coarse for a 2m/pix images. BTW, have you checked the SRTM, that is now provided globally at 30m/pix. In several areas it has proved to offer a less noisy DEM than GDEM.
1 - This is surprising. Can you check that the DEM does not contain any default missing values (such as -32000, -9999, etc.). If it does, you need to filter them out first
2 - Might be. Before correlation, just dynamic link your two orthos with ENVI, and blink between the two. They must be pretty well registered at that point. If it is not the case, then the ortho/registration part did not work well and need to be corrected first before going further
3 - Most probably something is wrong with the orthorectification mapping. Bilinear is a lower quality version of the Sinc, so it is not supposed to work better. Now, if the incidence angles are strong, then the orthorectified image should present some blurry/distorted parts when using the Sinc. This is normal and not artefact.

Francois

4

Re: spot 6

Hi dear Francois

I have two SPOT-6 data with 15.09 and 20.43 degrees incidence angles.
when I want to do ortho step these values of incidence angle cause bad results.
Is there a way to eliminate this defect, do you know?
Thank you in advance

5

Re: spot 6

Hi Anna,
the incidence angle by itself is not the cause of bad results.
It is not clear what you mean by bad results. Are you talking about the orthorectified image or the subsequent correlation of the orthorectified images?
With high incidence angle, and depending on the topography, a "bad" orthorectified image is usually due to poor topography correction (low resolution DEM).