Underlying topography extraction over forest areas from multi-baseline PolInSAR data

In this paper, the digital elevation model (dem) for a forest area is extracted from multi-baseline (mb) polarimetric interferometric synthetic aperture radar (polinsar) data. on the basis of the random-volume-over-ground (rvog) model, the weighted complex least-squares adjustment (wclsa) method is proposed for the ground phase estimation, so that the mb polinsar observations can be constrained by a generalized observation function and the observation contribution to the solution can be adjusted by a weighting strategy. a baseline length weighting strategy is then adopted to syncretize the dems estimated with the ground phases. the results of the simulated experiment undertaken in this study demonstrate that the wclsa method is sensitive to the number of redundant observations and can adjust the contributions of the different observations. we also applied the wclsa method to e-sar l- and p-band mb polinsar data from the krycklan river catchment in northern sweden. the results show that the two extracted dems are in close agreement with the light detection and ranging (lidar) dem, with root-mean-square errors of 3.54 and 3.16 m. the dem vertical error is correlated with the terrain slope and ground-cover condition, but not with the forest height.