A survey of temporal decorrelation from spaceborne L-Band repeat-pass InSAR

In this paper we quantify the effects of temporal decorrelation in repeat pass synthetic aperture radar interferometry (InSAR). Temporal decorrelation causes significant uncertainties in vegetation parameter estimates obtained using various InSAR techniques, which are desired on a global scale. Because of its stochastic nature temporal decorrelation is hard to model and isolate. In this paper we analyze temporal decorrelation statistically as observed in a large swath of SIR-C L-Band InSAR data collected over the eastern United States, with a repeat pass duration of one day in October 1994 and a near zero perpendicular baseline. The very small baseline for this particular pair makes the effect of volumetric scattering on correlation magnitude statistics nearly imperceptible, allowing for a quantitative analysis of temporal effects alone. The swath analyzed in this paper spans more than a million hectares of terrain comprised primarily of deciduous and evergreen forests, agricultural land, water and urban areas. The relationships of these different land-cover types, phenology and weather conditions (i.e. precipitation and wind) on the measures of interferometric correlation is analyzed in what amounts to be the most geographically extensive analysis of this phenomenon to date.

Research Highlights
► Temporal decorrelation is an error source for repeat-pass InSAR.
► One day, zero-baseline SIR-C L-band data was analyzed over a 1500 km swath.
► Temporal decorrelation is shown to be target and weather dependent.
► Probability density functions are provided for different land-cover classes.