The Double Interpolation and Double Prediction (didp) Approach for Insar and Gps Integration

The technique of synthetic aperture radar interferometry (or InSAR) was first suggested in 1974. Due to its characteristics of very high spatial resolution and wide coverage, as well as the cost effectiveness of the technique, it is not surprising that many earthquake ruptures have been studied using InSAR. However, InSAR is very sensitive to errors such as atmospheric effects, satellite orbit error, the condition of the ground surface and temporal decorrelation. Furthermore the repeat cycle of 35 (ERS-1) to 44 (JERS-1) days for SAR satellites cannot provide sufficient temporal resolution for some applications. On the other hand, many Continuous GPS (CGPS) networks of receivers have been established in recent years. CGPS can provide temporally dense observations (e.g. sampling at an interval of 30 seconds or higher), but these GPS receivers, spaced at about 25km apart for the densest arrays, are not spatially dense enough to characterise the dynamics of seismic faults. These require sub-kilometre level spatial resolution. Therefore, the two techniques of InSAR and CGPS appear to be very complementary. The proposed Double Interpolation and Double Prediction (DIDP) approach can be used to integrate InSAR and GPS results. The first step is to derive atmospheric corrections to the InSAR images from CGPS data, using techniques borrowed from 'GPS meteorology'. The second step is to remove or mitigate SAR satellite orbit errors by using the GPS results as constraints. These atmospheric and orbit corrections generate a GPS-corrected InSAR result. In the third step, the CGPS observations separated by one or several InSAR repeat cycles are densified in a grid manner by interpolation in the spatial domain, based on the GPS-corrected InSAR results; and then the densified grid observations are interpolated in the time domain using the daily, hourly, or even 30 second sampling rate of the CGPS time series. In the fourth step, based on the double interpolation result of the third step, forward filtering is used to predict the crustal deformation at all points on the grid. This is in fact a double prediction in both the temporal and spatial domains. Some algorithms and preliminary results of the DIDP technique for integrating InSAR and GPS results will be presented in this paper.