Much progress has been made in our ability to detect deformation on volcanoes using differential SAR interferometry (DInSAR), however significant problems with the technique remain. These include temporal decorrelation and the inability to interfere SAR images gathered on repeat satellite tracks more than a few hundred meters apart. One method for avoiding both temporal and spatial decorrelation is to identify persistent scatterers (PS) in the images. Their persistence means they do not decorrelate significantly with time, while their pointwise nature means that speckle is not an issue.
Other PS systems have been developed on the assumption that, to first order, the deformation is steady in time. While they can easily be adapted to include a non-steady model of deformation, they are limited to cases where the function of deformation with time can be simply parameterized, e.g a polynomial. Volcano deformation however, tends to be non-steady and not readily parameterized. We have therefore developed a system that generates a time series of deformation for each PS with no prior assumptions about the temporal nature of this deformation. An appropriate model of deformation can then be selected and solved for. This technique has been verified using synthetic data representing a non-steady, inflationary/deflationary Mogi source.