Phase Bias in short-interval interferograms: characteristics and mitigation strategies

Abstract:

Interferometric Synthetic Aperture Radar (InSAR) is a powerful tool for monitoring ground deformation. Until recently, the accuracy of estimated deformation had been assumed to depend only on a number of well-known phenomena, including atmospheric delays, topographic and orbital errors, unwrapping errors, and phase decorrelation due to changes in scattering behavior and system noise. However, recently an additional source of phase noise has been identified by Ansari et. al., which is strongest in short-interval multi-looked interferograms and, unlike other sources of noise, leads to biased, non-zero loop closure phases. This is problematic for time-series analysis because short-interval interferograms may be the only ones that maintain coherence for some areas. Here we explore the characteristics of this phase bias and discuss potential mitigation strategies.

To isolate the phase bias, we construct “daisy chain” sums of interferograms covering an identical 360 day time period, but using different time period interferograms (6,12,18, 24...180 days). Conventional noise sources in each of these sums should be nearly identical, and any differences should be small and centred on zero. However, we find that the daisy-chain sums that use short-time-interval interferograms are biased for some pixels, and that the bias appears to be spatially correlated. The bias is much weaker in longer-interval daisy-chain sums. We find that spatial filtering, in addition to multi-looking, significantly increases the bias. We investigate how the bias builds through time, and find that it does not build at a constant rate.

Additionally, we have performed a set of correlation analysis in order to find the physical cause of the phase bias. In agricultural areas, where the bias is strongest, there is a strong correlation between cumulative phase bias and cumulative values of integrated temperature gradients (growing degree days, GDD), especially in the case of the shortest temporal baselines. This suggests the phase bias may be related to vegetation growth, as postulated by Ansari et al.

We test several empirical approaches for correcting the bias using the differences in its magnitude between different daisy chains and using independent information such as GDD. We compare our empirical corrections with the EMI approach by Ansari et al.