Multi-baseline interferometric SAR for iterative height estimation

The tradeoff in choosing baseline length in interferometric SAR is that short baselines allow higher error while long baselines cause greater phase unwrapping problems. Combining information from baselines of different lengths can take advantage of the greater accuracy of long baselines while determining the phase wrapping from shorter baselines. This paper describes an iterative method which starts at the shortest baseline and proceeds to longer and longer baselines. The height is estimated for each baseline, using the previous estimate only to determine the phase wraps. If the signal-to-noise ratio of the interferometer is known and a value for the probability of error in the phase wraps is chosen, an optimal sequence of baseline lengths can be calculated. Simulations show that errors tend to propagate and become worse, so mean or median filters are used between each iteration. This paper will describe the technique and show examples from a simulated scene.