Projektbeschreibung (Englisch)
Numerical simulations and process analysis of CO2 enrichment in the vadose zone and at the gas-water interface of karstic systems
The vadose zone of many karstic systems shows strongly elevated CO2 concentrations in the gas phase, which can be attributed to seasonal microbiological activity dependent on soil moisture and temperature as well as the absence of stronger ventilations. CO2 has a higher density than air and tends to accumulate in deeper regions, though the time scales of such a potential gravity segregation process are subject to a number of parameters that are highly variable and hard to control in realistic field conditions. We can get access to data from at least one site in Slovenia where a particular situation of high CO2 concentrations in a hundreds of meters long dead-end cave passage is documented; there, the conditions of gravity segregation re-establish even after a mixing event during a short several-hours time-interval, which cannot be explained by the time-scales of gravity segregation alone, but requires instead the presence of highly conductive fractures and faults where CO2 may have had long time to accumulate.
The aim of the proposed study would include several steps, (1) to set up a generic conceptual model of the site, (2) to conduct a process analysis on the basis of a complex numerical model which would include a free flow (gas phase) in the cave passage coupled to a discrete fracture-matrix model to represent the karstic rock matrix. The fluid-system to be considered includes two-phase water-gas and three-component water, air, CO2. (3) Surrogate models, potentially with proxy physics need to be conceptualized which then allow for machine-learning of characteristic processes with the help of the available data (this includes that some of the data need be measured additionally according to the demands of the models).
A second site in Bavaria (Frankonian Jura) is envisaged for a potential measurement campaign to be conducted in order to collect data on CO2 enrichment at the gas-water interface. The site includes a ponor where water with enriched CO2 concentrations gets regularly diluted after heavy rain events. This site is a potential candidate to measure under in-situ conditions the enrichment of the water with CO2 due to density-driven dissolution at the gas-water interface during dry conditions (no flood events). The cave site is difficult to access and requires exploration and preliminary measurements before a potential larger campaign can be prepared.
A third aspect of the project relates to radon concentrations in caves, which seem to be correlated to CO2 concentrations. On the one hand, this correlation can be expected, since high concentrations of both gases are favored by low ventilation. On the other hand, the source terms for both gases have completely different origin. The setup of a time-resolved measurement of radon is planned at our field site in the Laichinger cave in the Swabian Jura.
Relation to the formulated RQs in the new SimTech proposal:
The overall vision for the large scale is to develop knowledge-integrated models (RQ1). We have here a situation, where parts of the effective karst-system-scale processes can be learned and upscaled, for example, from highly resolved fracture scale simulations (Keim & Class, 2024, submitted to WRR). On the other hand, we have data-rich regimes in the accessible regions of a karst system, i.e. the atmospheric conditions like temperature, precipitation as well as CO2 concentrations in caves. However, the vadose zone, where transport of CO2 and water occurs, is close to a black-box. We plan that machine-learning algorithms on the basis of the available data (and those still to be measured) can help to fill the gap in these vadose zone processes. For that reason, our strategy is to set up first generic high-fidelity models (with the available computational capabilities of the Dumux coupling interfaces) of the karst system which may later on in the potential new SimTech period be adapted to data-informed low-fidelity models (RQ2). RQ3, the development of interpretable geosystem models, is inherent to this endeavour.
Leiter
apl. Prof. Dr.-Ing. Holger Class
Bearbeiterin
Vivien Langhans (M.Sc.)
Laufzeit
09/2024 - 12/2025
Kontakte
Holger Class
apl. Prof. Dr.-Ing.Kommissarischer Leiter des Lehrstuhls
Vivien Langhans
Wissenschaftliche Mitarbeiterin