I am generating time series of 6-day coherence from S1A-S1B combinations since October. I find periods where I have consistently low coherence that is not due to terrain change or weather, but likely linked to baseline decorrelation. Is it possible to retrieve baselines for SLC pairs in SNAP and determine when these are least likely to be suitable for coherence analysis?
IMO it’s unlikely that baseline decorrelation is an issue with S1 unless the height-differences between nearby pixels are extreme (like in some urban or tropical forest scenarios). What makes you think weather (rain event, strong wind etc.) is not the cause? Small problems in InSAR-processing can also cause low coherence so you could check that the co-registration has performed well.
The area of interest are mostly bare and flat agricultural fields (Western Europe) and I control for the weather.
How do I determine that co-registration has performed well and/or that I have small problems in InSAR processing?
I use the standard processing sequence, which generally results in excellent coherence imagery, but not for particular pairs or sequences of pairs (in particular in the first half of February). So determining the baseline is still interesting as a causal parameter.
Is wet snow a possibility? That could cause very low coherence indeed. I haven’t personally used SNAP for InSAR in a long time so I don’t remember where the baseline-info is displayed. Co-registration quality parameters should be in some of the logs if I’m not mistaken.
To see the baselines prior to coregistration, you can use the InSAR Stack Overview in the InSAR menu. On a stack, you can use InSAR Stack Tool Window found in View -> Tool Windows -> Radar
Thanks for the tips. I get the following baselines:
For early February pairs (with overall low coherence results):
master slave Bperp
S1B 0203 S1A 0209 67.40
S1A 0206 S1B 0212 -102.73
S1B 0209 S1A 0215 91.98
and for March pairs (with overall high coherence results):
S1B 0311 S1A 0317 -88.66
S1A 0317 S1B 0323 50.16
so, I guess not drastically different and a possible cause of decorrelation. Would be nice if the InSAR Stack Tool function would be available in gpt.
I’ll dig a little further. There are frost/thaw cycles, but I had those in January as well, where coherence is not consistently low.