ON THE EFFECT OF REFERENCE FRAME MOTION ON INSAR DEFORMATION ESTIMA TES

For processing of interferometric synthetic aperture radar (InSAR) data, precise satellite orbits are required. These orbits are given in a reference frame with respect to which tectonic plates perform a relative motion. Neglecting this motion can cause temporally increasing baseline er- rors that induce large scale error ramps into the inter- ferometric phase. The amount of error depends on the geographical location and is evaluated globally for the ENVISAT orbit. Predicted biases of deformation esti- mates can reach up to 7 mm/a in some areas. Whereas these biases are not separable from actual deformation signals by spatio-temporal correlation properties, they are well predictable and can easily be accounted for. A most simple correction approach consists in compensating the plate motion by modifying orbital state vectors, assum- ing a homogeneous velocity for the whole plate. This approach has been tested on Persistent Scatterer Inter- ferometry (PSI) results over the area of Groningen, the Netherlands. InSAR deformation analysis is based on comparing the measured interferometric phase of two images with the reference phase computed from acquisition geometry. The latter is deduced from precise orbit ephemerides, which are commonly expressed in the International Ter- restrial Reference Frame (ITRF). This frame is a realisa- tion of a global coordinate system and defined by a num- ber of geodetic stations close to the earth surface. To ac- count for secular tectonic motion, not only positions but also linear velocities are attributed to the ITRF stations. A common assumption for InSAR processing is that the reference system of the orbit data does not move with re- spect to the earth surface. However, due to plate tecton- ics, this assumption is not valid at the centimetre level. Observing from a viewpoint on a tectonic plate, the co- ordinate frame of the orbits performs a relative motion in the order of centimetres per year. Neglecting this in In- SAR processing is comparable to making an error in the interferometric baseline, the size of which is increasing with the temporal baseline. The effect on the interfero- metric phase is an almost linear trend in range, suggest- ing a large scale tilt of the surface. In contrast to tem- porally uncorrelated ramps caused by orbital errors, this kind of trend is correlated in time and cannot be sepa- rated from spatio-temporally correlated deformation sig- nals. Hence, it can mistakenly be interpreted as an actual deformation signal in applications where the signal of in- terest is a large scale deformation over a long time period, e. g., in monitoring interseismic tectonic motions. This contribution investigates the effect of neglecting the relative tectonic motion on InSAR deformation estimates, henceforth referred to as the reference frame effect. Af- ter a brief review on the ITRF and orbit errors in SAR interferometry, the effect will be described in detail and evaluated globally. A simple correction approach will be proposed and applied on ENVISAT PSI results from the Groningen area in the Netherlands.