On the Use of Point Target Characteristics in the Estimation of Low Subsidence Rates Due to Gas Extraction in Groningen, the Netherlands

The subsurface of the north-eastern part of the Netherlands contains large gas reservoirs that have been taken into production since the early 1960s. The resulting subsidence has been precisely and reliably estimated using measurements from repeated leveling campaigns and is now for the first time unambiguously observed using PS-InSAR. The subsidence rates are small (< 1 cm/year) and the PS density is relatively low due to the rural character of the area. Particularly it is difficult to distinguish the PS deformation components due to gas extraction from other possibly superposed deformation regimes. Besides a strict quality description of the estimated parameters, the characterization of physical PS properties is therefore of major importance. In this paper we focus on the interpretation of PS-InSAR deformation estimates in Groningen by investigating the scattering properties of the detected PS. Through combining the shape of their complex spectrum and estimated heights from PS-InSAR processing, the nature of the PS is investigated. Important issues for performing the scatterer characteristics analysis, such as coregistration and SAR amplitude calibration, are addressed. After analyzing the effect of temporal and perpendicular baseline and Doppler frequency variations, the PSInSAR deformation estimates from different reflection types are evaluated on their performance for estimating subsidence due to gas extraction. 1 THE GRONINGEN GAS FIELD The north-eastern part of the Netherlands contains large gas reservoirs at a depth of approximately 3 kilometers below ground level. The largest is the Groningen gas field with a diameter of roughly 30 kilometers. The reservoir itself consists of a 100-200 m thick porous sandstone layer. Since the early 1960s it has been taken into production by the Nederlandse Aardolie Maatschappij B.V. (NAM). During the production, the reservoir pore pressure is decreasing resulting in a compacting layer. As the layers on top of the compacting reservoirs are subsiding as well, the ground level is subsiding. Surface deformation due to gas extraction has the shape of a spatially smooth ellipsoidal subsidence bowl [1], depending on the depth and shape of the reservoir. The subsidence rate is approximately linear, with a decreasing velocity further away from the center of the bowl. The NAM is legally obliged to monitor the subsidence due to gas extraction, to be able to assess the influence of gas extraction on water management and environmental issues. Hence leveling campaigns have been performed since the start of production. Geodetic adjustment and testing techniques have been applied to estimate subsidence due to gas extraction from height difference observations. Besides tracing measurement errors and point misidentifications also tests on autonomous benchmark behavior are evaluated. The total estimated subsidence in the center of the bowl is 24.5 cm for the period up to 2003, which implies an average subsidence rate of 7.5 mm/year. Currently, the application of PS-InSAR for estimating subsidence due to gas extraction is evaluated. Complicating factors are the low subsidence rates (< 1cm/year), the rural character of the area and atmospheric disturbances. Besides these issues, a major challenge lies in the interpretation of the estimated PS velocities. Contrary to traditional geodetic techniques using well-defined benchmarks, the PS measurement point – the effective reflectivity center – is less well known. As a result it may be difficult to discriminate which PS displacements are caused by a certain deformation regime (gas extraction, shallow compaction, structural instabilities). A way to gain more insight on PS properties, is to investigate if they can be classified in specular and dihedral reflections. With the objective to be able to discriminate subsidence due to gas extraction from other deformation regimes, the PS reflectivity patterns are analyzed. Methods as described in [2] are evaluated for the Groningen gas field. _____________________________________________________________ Proc. Fringe 2005 Workshop, Frascati, Italy, 28 November – 2 December 2005 (ESA SP-610, February 2006) 2 GRONINGEN PS-INSAR RESULTS 2.1 Processing methodology A PS-InSAR analysis has been performed both for ascending and descending mode. The coregistration residuals have been checked and appeared to be lower than 0.1 pixel for the majority of the acquisitions within the stacks. Ambiguities have been resolved using the ambiguity function, followed by a geodetic spatial network testing which should assure a success rate close to 1 for correctly resolving the ambiguities. The PS displacements are parameterized with a minimum number of parameters, which means linear velocity. This is the strongest model in terms of redundancy. It needs to be followed by a residual analysis to trace possible model deviations which influence the parameter estimates. Acquisitions with a Doppler difference (relative to master) higher than 500 Hz are excluded from the PS velocity and height estimation, as the noise level of most PS observations is significantly higher. Only few PS have such ideal point target properties, that they are visible over a wide range of squint angles. In the reflectivity analysis they are again included, as its strength depends on the variety of viewing angles in range and azimuth direction. The temporal acquisition frequency of the descending mode is twice the temporal frequency of the ascending mode. The number of interferograms used for the PS velocity estimation is 72 and 32 for descending and ascending mode, respectively. 2.2 PS-InSAR results The results in Fig. 1 show that PS-InSAR is able to estimate wide scale surface deformation coinciding with the gas extraction areas. The main Groningen subsidence bowl is clearly visible, as well as some other subsiding areas which are not necessarily caused by gas extraction. Subsidence rates vary generally from -8 mm/year up to 2 mm/year uplift relative to the reference point. They correspond with the subsidence rates estimated from the leveling campaigns, although this comparison has to be further refined. The density in the rural areas is remarkably high. Although the Groningen area contains numerous agricultural fields, the majority of the buildings in this area serves as a PS, especially when combining different tracks. The PS density is not comparable to that of an urban area (> 100 PS/km), but it is sufficient to cover the spatial extent of the subsidence bowl. The PS results of both ascending and descending mode detect the spatial subsidence pattern located at the Groningen gas field. After converting the PS velocities to the vertical, differences in average PS velocity per km remain with a standard deviation of 1.8 mm/year. First of all this may be due to different physical PS properties representing different deformation regimes. Especially in areas with a low PS density this may lead to local differences between ascending and descending velocities. Secondly, the lower observational redundancy in ascending mode (less images) may lead to less precise estimates.