Multi-baseline coherence optimisation in partial and compact polarimetric modes

Modern space-borne SAR sensors, like ALOS-PALSAR, TerraSAR-X and Radarsat-2 all provide at least a "partial polarimetric mode", acquiring only 2 of the 4 elements of the Sinclair matrix, like for example HH and HV or VV and VH. In addition, it has been demonstrated that with certain so-called "compact PolSAR" single-transmit dual-receive techniques one can obtain an estimation of the fully polarimetric information. Such systems are attractive in terms of reduction of pulse repetition frequency, data rate, and complexity and are currently very popular. However, they do not acquire complete information pertaining to the full polarisation state of the target and, as a consequence also coherence optimisation suffers from the reduced configuration space. In this paper, the potential of the different partial polarimetric setups for coherence optimisation is evaluated both theoretically and experimentally and compared to the capabilities of a fully polarimetric system. It will be analysed to which extent partial polarimetric system can improve the derivation of interferometric information from partly decorrelated surfaces, in particular of vegetated or even forested areas. Special attention is paid to the constrained coherence optimisation of multi-baseline setups, important for modern DInSAR techniques like PS analysis and continuous DInSAR monitoring in general. All experimental analyses will be performed using fully polarimetric multi-baseline data sets. For proper comparison, partial polarimetric information is derived from these by matrix transformations according to the respective transmit / receive configuration.