Abstract. Tunnelling below water passages is a challenging task in terms of planning, pre-investigation and construction. Fracture zones in the underlying bedrock lead to low rock quality and thus reduced stability. For natural reasons, they tend to be more frequent at water passages. Ground investigations that provide information on the subsurface are necessary prior to the construction phase, but these can be logistically difficult. Geophysics can help close the gaps between local point information by producing subsurface images. An approach that combines seismic refraction tomography and electrical resistivity tomography has been tested at the Äspö Hard Rock Laboratory (HRL). The aim was to detect fracture zones in a well-known but logistically challenging area from a measuring perspective.The presented surveys cover a water passage along part of a tunnel that connects surface facilities with an underground test laboratory. The tunnel is approximately 100 m below and 20 m east of the survey line and gives evidence for one major and several minor fracture zones. The geological and general test site conditions, e.g. with strong power line noise from the nearby nuclear power plant, are challenging for geophysical measurements. Co-located positions for seismic and ERT sensors and source positions are used on the 450 m underwater section of the 700 m profile. Because of a large transition zone that appeared in the ERT result and the missing coverage of the seismic data, fracture zones at the southern and northern parts of the underwater passage cannot be detected by separated inversion. Synthetic studies show that significant three-dimensional (3-D) artefacts occur in the ERT model that even exceed the positioning errors of underwater electrodes. The model coverage is closely connected to the resolution and can be used to display the model uncertainty by introducing thresholds to fade-out regions of medium and low resolution. A structural coupling cooperative inversion approach is able to image the northern fracture zone successfully. In addition, previously unknown sedimentary deposits with a significantly large thickness are detected in the otherwise unusually well-documented geological environment. The results significantly improve the imaging of some geologic features, which would have been undetected or misinterpreted otherwise, and combines the images by means of cluster analysis into a conceptual subsurface model.

The pore-size distribution (PSD) of geologic materials is an important rock parameter to understand the flow of water in the subsurface. PSDs can be obtained from sieving analyses, mercury porosimetry measurements, and imaging techniques, but none of these methods is available for in situ measurements. Nuclear magnetic resonance (NMR) measurements are controlled by rock parameters such as the surface-area to porevolume ratio. NMR is available for in situ measurements. State-of-the-art NMR relaxation time measurements need a calibration of the surface relaxivity ρ to extract pore-size information. State-of-the-art NMR diffusion measurements avoid the calibration of ρ but are limited to small pores. We developed an approach that estimates the average pore size without calibrating ρ by means of incorporating higher order modes into the signal interpretation of NMR relaxation times. We conducted forward-modeling studies using an analytic solution for cylindrical tubes, 2D finite-element simulations to incorporate fractal pore spaces, and laboratory experiments on synthetic and natural samples. Our experimental data indicated that relaxation can occur outside the fast-diffusion regime not only for coarse-grained materials, but also for fine-to medium-grained unconsolidated sandy materials due to high surface relaxivities. We found that the rock-fluid interface's roughness had a significant impact on the diffusion regime and led to an apparent increase in ρ, which may cause intermediate or slow diffusion. The methodology was limited to materials with a narrow PSD and uniform distribution of ρ because we assumed multiexponential decay due to diffusion in single isolated pores.

Electrical resistivity tomography (ERT) and refraction seismics are among the most frequently used geophysical methods for site-investigations and the combined results can be very helpful to fill in the gaps between the point measurements made by traditional geotechnical methods such as Cone Penetration Test (CPT), core-drilling and geophysical borehole logging. The interpretation of the results from a geophysical investigation constituting a single method often yields ambiguous results. Hence, an approach utilizing multiple techniques is often necessary. To facilitate interpretation of such a combined dataset, we propose a more controlled and objective approach and present a method for a structurally coupled inversion of 2D electrical resistivity and refraction seismic data using unstructured meshes. Mean shift clustering is used to combine the two images and to compare the separate and coupled inversion methodologies. Two synthetic examples are used to demonstrate the method, and a field-data example is included as a proof of concept. In all cases a significant improvement by the coupling is visible. The methodology can be used as a tool for improved data interpretation and for obtaining a more comprehensive and complete picture of the subsurface by combining geophysical methods.

Electric resistivity tomography (ERT) is a widespread technique used for geologic and hydrological subsurface characterization. The 3D application is limited to small scales due to the enormous effort required for field surveys. Long-electrode (LE)-ERT using steel-cased boreholes can overcome small-scale limitations and has recently gained significant attention. However, no systematic investigation concerning the performance of long electrodes has been performed. We have conducted synthetic studies of LE-ERT to understand its advantages and limitations. We have compared three different approaches of modeling long electrodes, placing particular emphasis on the complete electrode model, to determine which reflects the physical reality best and thus can be used as a benchmark. The conductive cell model led to comparable results but was numerically more expensive. An interesting alternative approach was the shunt electrode model, which does not require discretizing the lateral extension of the casing. Systematic sensitivity studies revealed that model resolution could be enhanced in the borehole depth range through the use of electrodes of different lengths or surface electrodes. Varying contact impedances along different sections of the borehole can change the electric field visibly and influence four-point measurements. Even if large electrode segments show reduced coupling, the influence on voltage measurements was below 4% for realistic contact impedances. We have proven the applicability of LE-ERT for imaging lateral saltwater movement through the simulation of simple scenarios in the context of saltwater intrusion. Our synthetic examples determined the advantage of long electrodes compared with simulations using surface electrodes. The 3D inversion of synthetic data sets revealed that the modeled anomalies could be imaged in most cases. Different models revealed limitations due to poor vertical resolution and supported the usage of casings with different lengths or in combination with surface electrodes to improve resolution.

Underground constructions for public traffic purposes are becoming increasingly important for urban areas in order to use the limited space more efficiently. Several electric resistivity tomography and seismic refraction tomography measurements were performed crossing a water passage near Stockholm during the pre‐investigation phase of a tunnel building project. The objective was to determine the bedrock interface and qualitatively assess the rock quality. The scope of this study is to present a field case in an urban environment and show improvements of geophysical results due to additional model constraints by a joint inversion. Results of individual inversions show a large transition zone below the seabed from electric resistivity tomography. Some parts of the seismic refraction tomography have a low model resolution, due to gas‐bearing sediments with a low velocity together with a high noise level, which leads to insufficient investigation depth that makes it difficult to determine the bedrock interface. However, the bedrock interface could be reconstructed in the resistivity model by performing a joint inversion, using the seismic velocity model to constrain the electric resistivity tomography model and vice versa. Adjacent geotechnical soundings support the joint inversion results. A vertical low resistive zone could be identified as a dominant fracture zone in the southern part of the investigated area. In general, the joint inversion approach significantly improved the electric resistivity tomography results and provided a more reliable bedrock estimation.