DEM generation and time series analysis of InSAR using Kalman filters

The elevation and movement history of radar scatterers can be obtained using interferometric synthetic aperture radar (InSAR) data. For any coherent imaging radar, the phase measurements for scatterers are expressed modulo 2π, causing ambiguity in the measurements. The phase measurements therefore must be “unwrapped” over all the data dimensions. Each synthetic aperture radar (SAR) image is collected in the two spatial dimensions of radar geometry: azimuth and range. Multiple images acquired from the same orbit at different times create a stack of SAR images, where the phase values in the stack constitute the third dimension: time. Solving the phase ambiguity of a point in this three-dimensional (3-D) dataset can be accomplished more accurately by using data from neighboring points in all three dimensions. Compared to unwrapping one - or two-dimensional subsets of the dataset, 3-D unwrapping is more complex and generally necessitates a combination of one - and two-dimensional filters. In this paper, we describe a novel 3D unwrapping algorithm based on Kalman filtering, apply the algorithm to DEM generation, and outline preliminary results of the time series analysis.