Optimum Interferogram Stack vs PS: Multifrequency Analysis

With the next generation of satellites, new possibilities are opened for SAR interferometry. Better orbital control and shorter revisiting times will allow to infer accurate information from targets usually considered unsuitable for multipass interferometric applications, such as vegetated areas. The usual PS-based techniques analyze phase differences with respect to a master acquisition. In order to exploit distributed targets, instead, all the available interferograms should be used, properly weighted. With a suitable model, we show how to find the weights in order to achieve the best estimates of DEM elevation and subsidence velocity, and their minimal dispersions. We consider different parameters, such as number of acquisitions, set of baselines, revisiting time intervals, size of the observed area, and carrier frequency. The target is modeled as being normally distributed, but the covariance matrix among the different acquisitions can cover the cases of decorrelating targets as well as PSs. Furthermore, we consider the presence of additive, spatially correlated, phase noise to represent the impact of the atmospheric phase screen. We compare the results for distributed targets to those of the PS technique in terms of spatial resolution and estimate accuracy. The performances achievable are predicted as a function of the carrier frequency. For this, we use a frequency dependent target decorrelation model, inferred from an analysis of the ERS 1 data (Rome, three days repeat cycle), together with a brownian motion model for the temporal changes of the scatterer.