Unwrapping the phase in InSAR data by a multibaseline approach. Test on a synthetic scene

Synthetic Aperture Radar (SAR) nowadays is used by a broad community for very different applications. Especially for cartography, data with high resolution in all three dimensions are requested to replace the time consuming manual mapping by a semi-automatic or automatic method. InSAR-Systems have the capability to provide the users with a better view of our planet's surface and even of that of other planets in all three dimensions. To gather data with good height estimation it is necessary to use a wide interferometric baseline or, if that is not possible, to go to higher frequencies than generally used by SAR-systems. This is because the height estimation is proportional to the radar frequency and reciprocally proportional to the interferometric baselength. However the problem of resolving the phase ambiguity at smaller wavelengths is more critical than at longer wavelengths, as the unambiguous height interval is also reduced by the quotient of wavelengths. But as the antenna diameter at equal gain is much smaller at higher frequency, it is easily possible to build up a multiple baseline antenna with a number of neighbouring horns and an increasing baselength going from narrow to wide. The narrowest, corresponding to adjacent horns, is then assumed to be unambiguous in phase. If this is not the case an unwrapping method as known from classical InSAR, such as the Goldstein- or Dipole-method, has to be used to make the phase unambiguous. This coarse interferogram, with low height estimation precision but fully unambiguous, is used as a starting point for the algorithm, which in the next step unwraps the interferogram with the next wider baseline using the coarse height information to solve the phase ambiguities. The result is a refined unambiguous interferogram, which is now used as new reference for unwrapping the interferogram belonging to the next, even wider baseline. This process is repeated consecutively until the interferogram with highest precision, belonging to the widest baseline, is unwrapped. On the expense of this multi-channel-approach the algorithm is relatively simple and robust, and even the amount of processing time is reduced, compared to other competing methods. In this paper we describe the algorithm and show the results of tests on a synthetic area as an example. Possibilities and limitations of this approach are discussed.