SHEAR WAVE VELOCITY FOR THE UPPER 30 M: COMBINING A 3D VOXEL MODEL AND SEISMIC CPTS FOR THE GRONINGEN GAS FIELD, THE NETHERLANDS

The Groningen gas field in the Netherlands is one of the largest gas fields of Europe and has been in production since the 60’s. Due to the progressive depletion of the reservoir, induced seismic activity has increased in recent years. In 2012, an earthquake of magnitude 3.6 initiated further research in prediction and management of risks related to man-induced earthquakes.

The work reported here deals with the derivation of spatially distributed Shear Wave Velocity (Vs) for the upper 30 m of the subsurface column (Vs₃₀). The Geological Survey of the Netherlands and Deltares combined a beta version of the GeoTOP model of the area and seismic cone penetration tests (SECPT) into a Vs₃₀ model of the area covering the gas field. The GeoTOP model is a 3D voxel model, in which each voxel is attributed with lithostratigraphy and lithological classes (peat, clay, fine sand, etc.). The modelling procedure of GeoTOP starts with the modelling of top and bottom of each stratigraphical unit. Next, the 3D volume in between top and bottom (voxels) is filled with lithoclasses using geostatistical indicator simulation routines. This results in a 3D model in which each voxel has attributes describing the lithostratigraphial unit and the most likely lithoclass.

60 SECPT’s were used to derive statistical distributions (with mean and standard deviation) of Vs for each combination of lithostratigraphical unit and lithoclass. In this way it was possible to assign a specific Vs to each voxel in the model.

For each voxel in the stack of voxels that covers the upper 30 m (i.e. 60 voxels), a Vs value was randomly drawn from the statistical distribution of the lithostratigraphical – lithoclass combination it belongs to. The Vs₃₀ for each voxelstack is then calculated using the harmonic mean of the Vs of the 60 voxels. By repeating this procedure 100 times, the uncertainty in Vs₃₀ was determined.

Using the above described procedure we were able to delineate zones with distinct Vs₃₀ characteristics: areas containing predominantly soft Holocene deposits with low Vs₃₀ and areas with predominantly stiff Pleistocene deposits with high Vs₃₀. Also the uncertainty in Vs₃₀ could be quantified. This is a huge improvement compared to the previously used Vs₃₀, which was one value for the entire gas field. The Vs₃₀ will be used as input for site amplification predictions.