Towards the Fusion of GNSS and InSAR Observations for the Purpose of Water Vapor Retrieval

Incompletely or incorrectly modeled atmospheric effects limit the quality of the exploitation of observations of space-based geodetic sensors, such as GNSS (Global Navigation Satellite Systems) and InSAR (Interferometric Synthetic Aperture Radar). In contrast, state variables of the Earth’s atmosphere, especially water vapor, contain valuable information for climate research and weather forecasting. The Institute of Photogrammetry and Remote Sensing (IPF) and the Geodetic Institute (GIK) of the Karlsruhe Institute of Technology (KIT) carried out various research for atmospheric water vapor retrieval. For further investigations, we focus on the quality of water vapor estimates from the geodetic sensors GNSS and InSAR. Surface meteorological information is taken into account for the described analysis. Data from the MEdium Resolution Imaging Spectrometer (MERIS) are used for validating our estimates. The area under investigation is the Upper Rhine Graben (URG), which is covered by the dense GNSS network GURN (GNSS Upper Rhine Graben Network) since 2002. A stack of 17 Envisat SAR acquisitions was available. These SAR data cover a 100 km× 100 km region in the URG. The SAR images were acquired between 2003 and 2008, but most of them are concentrated in the year 2005. The described project aims at a straightforward comparison of the wet delay, caused by water vapor, derived from GNSS and InSAR. Therefore, the InSAR neutrospheric phase has to be separated from other components contained in InSAR measurements. For this purpose, it is assumed that the surface deformation within the area under investigation is negligible during the considered period of time. In the case of InSAR, persistent scatterer interferometry is used and leads to the observation of differential wet delays. Within the GNSS data processing, the Precise Point Positioning (PPP) approach is applied to estimate the total neutrospheric path delay. The total path delay deduced from GNSS is composed of a prediction model, estimated site-specific neutrosphere parameters (SSNP), and horizontal neutrospheric gradients as well as observation residuals. Based on an additional, comparative GNSS study carried out with respect to the predicting Niell mapping function (NMF), the effect of the weather model based Vienna mapping function (VMF) on the GNSS results is evaluated. Most important findings of the GNSS-related research of this work are: The SSNP deduced by means of the VMF attain smaller values than those derived from NMF. However, their effect on the total wet delay is significant and they may not be neglected. On the contrary, independent of the mapping function, the effect of the estimated horizontal gradients deduced from observations down to 3◦ elevation is classified as insignificant with regard to path delays observed at elevation angles above 45◦ (InSAR elevation angle: about 65.5◦). However, the phase residuals contribution to the satellite-directed path delays is very important. The annual standard deviations of the site height components determined within the GNSS data processing based on NMF have larger values than those based on the VMF. In contrast, the site latitudes and longitudes based on the NMF and VMF remain unchanged at the representative sample of GNSS sites. Comparisons of GNSS observations with the satellite-directed InSAR data show that only a partial component of the wet delay remains after the interferogram formation. A comparison with partial neutrospheric delays deduced from MERIS shows strong agreement between InSAR and MERIS data. This implies that a topography-dependent component as well as a linear trend contained in the wet delay from MERIS or GNSS have to be reduced to emulate the partial wet delays from InSAR.