ALBERT

Description: The general circulation of the world ocean generates characteristic magnetic signals by interacting with the ambient geomagnetic field. These ocean-induced magnetic signals can principally be measured by satellites and could serve as indirect observations of the ocean. Since the so-called motionally induced magnetic field is to first order proportional to conductivity-weighted and depth-integrated ocean velocities, global oceanic magnetic field observations could provide new constraints on oceanic transports of water, heat, and salinity. However, many aspects of electromagnetic induction in the ocean are either not well understood or unknown. This ranges from the basic characterization of motional induction in the ocean to possible applications and benefits for ocean modelling. This cumulative thesis encompasses the characterization of electromagnetic induction in the ocean, both in terms of physical properties and model uncertainties. One new application of electromagnetic induction in the ocean is investigated, namely the possibility to constrain an ocean general circulation model with satellite observations of the ocean-induced magnetic field. An electromagnetic induction model is implemented into an ocean general circulation model. This model combination allows the investigation of specific influences of seawater properties on motional induction. In previous studies, the electric conductivity of seawater was often treated in a simplified way by assuming it to be uniformly distributed in the ocean and temporally constant. In the first application of the combined numerical models, it is shown that this assumption is insufficient for capturing the temporal variability of motional induction accurately. Considering a realistic three-dimensional seawater conductivity distribution based on ocean temperature and salinity increases the temporal variability of ocean-induced magnetic signals by up to 45 %. These changes are found to predominantly originate from large vertical gradients of seawater conductivity in the upper ocean. The modelling of the general ocean circulation and of motional induction is affected by various uncertainties and errors, which are introduced by forcing input data and by the numerical models themselves. For potential applications of motional induction, e.g., a reliable comparison of model results with observational data, or data assimilation experiments, a realistic estimation of model uncertainties is essential. Ensemble simulations based on different error scenarios are performed to estimate the aggregated uncertainty of the modelled ocean-induced magnetic field. It is shown that the uncertainty of the modelled ocean-induced magnetic field reaches up to 30 % of the signal strength and is subject to large spatial and seasonal variations. The wind stress forcing of the ocean model is a major source of uncertainty. However, specific spatially and temporally robust regions are identified in the ocean-induced magnetic field that retain a small uncertainty in all error scenarios. Based on the previous findings, data assimilation experiments with artificial satellite observations of the ocean-induced magnetic field are designed and conducted for the first time. In a model-based twin study, artificial satellite observations of the oceanic magnetic field are generated and sequentially assimilated into an ocean general circulation model with a localized ensemble Kalman filter. The impact of the data assimilation on the induced magnetic field, ocean velocities, temperature, and salinity is measured. Compared to a reference simulation without data assimilation, the ocean-induced magnetic field is improved by up to 17 % globally and up to 54 % locally. Improvements of the underlying depth-integrated ocean velocities show values up to 7 % globally, and up to 50 % locally. These improvements result from a consistently better recovery of ocean velocities from the sea surface down to the bottom of the ocean. However, the Kalman filter fails to improve ocean temperature and salinity globally. Kalman filter adjustments of the wind stress forcing of the ocean model are found to be essential for a successful data assimilation.

Description: Carbon Capture and Storage (CCS) is considered as a promising measure to reduce anthropogenic greenhouse gas emissions into the atmosphere. Scientific assessments suggest that deep porous rock formations saturated with brine (saline aquifers) provide the largest storage potential due to their abundance in the Earth’s sedimentary basins. However, geological underground storage of CO2 (carbon dioxide) may also cause serious negative environmental and infrastructural impacts. The far-reaching pressure build-up affects regional fluid flow and may compromise mechanical rock integrity by changes in the recent stress field. Structural failure of reservoir, caprock or adjacent fault zones, accompanied by CO2 leakage or large-scale displacement of brines are among potential risks associated with CO2 injection into deeper saline formations. If brine reaches shallower aquifer complexes by upward migration through conductive pathways such as improperly sealed abandoned wells, permeable faults or erosive discontinuities in the overlying rocks, freshwater resources can be endangered by salinization. The present thesis aims at evaluating the hydraulic and mechanical impacts of industrial-scale CO2 injection and the resulting pressure increase for a potential saline onshore storage formation in the Middle Buntsandstein sequence of the Northeast German Basin. Here, the main emphasis is to assess the degree and bandwidth of potential shallow aquifer salinization by upward brine migration through permeable regional fault zones at the Beeskow-Birkholz storage site. Thereby, it is important to determine, which geological conditions promote upward brine displacement in geological CO2 storage, and whether the pressure build-up affects the mechanical integrity of fault zones and/or caprocks. Four geological 3D models with an extent between 1,765 km2 and 10,000 km2 and different layer structures serve as the basis for this research and are implemented in multi-phase flow and coupled hydro-mechanical simulations. The methodology applied for model set up and data integration varied, depended on the respective focus of investigation. For flow simulations, regional fault zones are described in the models either by real grid elements or by virtual elements that allow for a discrete fault representation without introducing specific grid refinements in the near-fault area. In the mechanical simulations, a plasto-elastic constitutive model for fault zones is applied, using embedded weak planes of corresponding dip angle and dip direction at the respective fault element locations. Multi-phase flow simulation results show that the magnitude of pressure build-up in the storage formation and pressure development over time determine the intensity and duration of brine flow into overlying aquifers. Salinity in the shallower aquifer increases only locally close to the fault zones, whereby the degree in salinization mainly depends on the lateral boundary conditions, the effective damage zone volume of fault zones, the presence of overlying reservoirs and the initial salinity distribution defined for the simulation scenario. The permeability of fault zones, however, has a comparatively minor impact on shallow aquifer salinization. Short hydraulically conductive fault segments lead to the highest local salinity increase, whereas laterally open boundaries and overlying reservoirs connected to the fault zones significantly diminish the risk of shallow aquifer salinization. Knowledge on the initial salinity distribution in the fault is essential for salinization assessments, since the displaced brine originates from the upper part of the faults only, and not from greater depths. Structural failure of fault zones as a consequence of injection-induced pressure build-up and effective stress changes can increase the risk of upward brine migration. To assess the fault slip and dilation tendencies at the respective site, one-way coupled hydro-mechanical simulations were applied in subsequent analysis. A one-way coupling procedure considers the time-dependent pore pressure development obtained from the dynamic flow simulations as input to hydro-mechanical simulations. The hydro-mechanical simulator then calculates potential rock mass failure resulting from stress changes without providing feedback to the flow simulator. In a first approach, the pressure distribution obtained from the flow simulations, was fitted by polynomial functions and integrated into the hydro-mechanical simulator for selected time steps. Simulation results demonstrate that only very few fault elements in the model are affected by shear and tensile rock failure, so that the development of a consistent slip plane along the faults, and thus fault reactivation is consequently expected not to occur at the Beeskow-Birkholz site under the given assumptions. For coupling evaluation, the applied one-way procedure was carried out for an equivalent saline onshore storage site in the Norwegian-Danish Basin close to the city of Vedsted, including a numerical modelling benchmark against the results produced by another well-established modelling group. The application of identical models for this process coupling allows an element-wise implementation of the time-dependant pore pressure distribution from the dynamic flow into the hydro-mechanical simulator. Simulation results show that mechanical impacts are mainly determined by fault conductivity and caprock permeability, which are influencing the spatial pore pressure distribution. A higher permeability of the caprock above the storage formation consequently induces higher vertical uplifts at the ground surface. In the present thesis, it is shown that the presence of hydraulically conductive faults must not necessarily lead to shallow aquifer salinization, since various factors have been proven to influence the occurrence and degree of salinization under the tested constraints at the respective storage site. The magnitude of pressure increase in the reservoir is the driving factor in upward brine migration through fault zones: larger pressures induce stronger brine displacement, and consequently result in higher salinities in shallow aquifers. The magnitude of pressure build-up in turn, depends on the chosen lateral boundary conditions, the presence of overlying reservoirs and the effective damage zone volume of faults. At Beeskow-Birkholz, shallow aquifer salinization did not occur over large areas and the faults were not affected by structural failure. However, if brine reaches groundwater bodies, the local maximum salinity increase above the salt-freshwater boundary can reach a concentration larger than the limit prescribed by the German Drinking Water Directive. In summary, numerical models can be well applied to obtain site-specific insights into the fluid flow dynamics in geological CO2 storage. At the same time, the simulations help to identify the geological conditions with the greatest impact on upward brine migration and provide an initial assessment of the anticipated risks including their extent and significance.