For an unexpected reason I had to run the coregistration and coherence estimation of two SLC IW Sentinel1 products with VV polarization. They are two spatial and temporal consecutive frames ( red and blue swaths in the image). Probably it is not such a meaningful operation but I had to look into it and I need some explanation about the results.
The coregistration has been performed after typical preprocessing:
Apply orbit
calibration
topSAR split
and then starting with the subswath:
back-geocoding
topSAR deburst
topSAR merge
The result of the coregistration was the intensity of the master product, but the intensity of the slave contains only NO-DATA pixels.
when computing the coherence ( after the back-geocoding for each subswath)
the final result is unexpectedly a matrix with all valid values.
So how these two behaviours can be explained for the same pair of products?? I supposed the coherence would had been with only NO-DATA values.
Help me @ABraun@lveci@mengdahl@marpet
It’s not clear what you have attempted. You need two acquisitions from the same track , taken at different time-periods (time difference is multiple of 6/12 days).
@mengdahl
The temporal baselines are
ALos2 before pair - 99 days
ALos2 after pair - 57 days
Sentinel 1 before pair - 12 days
Sentinel 2 after pair - 120 days
@mengdahl the closest image for S1 is the 120 days.
is there a co-registration parameter that can be tuned to decrease the effect of large temporal baseline.
The acquisition scenario must have changed, a pity. Are you sure there are no images from another track for example? The temporal decorrelation cannot be compensated for in processing as the actual scatterers on the ground have changed/moved.