Geochemical modeling of CO2–water–rock interactions for two different hydrochemical types of CO2-rich springs in Kangwon District, Korea

• CO2-rich springs as a natural analogue of the seepage of sequestered CO2.
• Ca-HCO3- and Na-HCO3-type springs occur together in a granitic terrain.
• Geochemical modeling explained the observed hydrochemical changes.
• Na-HCO3-type springs are formed by silicate weathering under high PCO2 condition.
• Ca-HCO3-type springs by reverse cation exchange during shallow water mixing.

Naturally outflowing CO2-rich springs are a natural analogue of the seepage of sequestered CO2 in geological storage sites. In Kangwon district of South Korea, two hydrochemically different types of CO2-rich springs (i.e., Ca–HCO3-type and Na–HCO3-type) occur together in a granitic terrain. Hydrochemical and water-isotope data (i.e., δ18O–δD and tritium) show that Na–HCO3-type springs have experienced significant silicate weathering processes over a long residence time at depths, while Ca–HCO3-type springs were formed by the mixing of Na–HCO3-type springs with shallow groundwater during ascent. In this study, diverse geochemical models including mixing, ion exchange and reaction path were investigated to verify the geochemical processes accounting for the occurrence of two contrasting types of CO2-rich springs. The mixing and ion exchange models reveal that Ca–HCO3-type springs are well explained by reverse cation exchange occurring during the mixing of Na–HCO3-type springs with shallow groundwater. The Na–HCO3-type springs are well explained by the reaction path modeling including the dissolution of silicate minerals (plagioclase, K-feldspar and biotite) and the precipitation of secondary minerals (calcite, kaolinite, muscovite and Mg-beidellite), implying that dissolved carbon is sequestered by calcite precipitation (i.e., mineral trapping). However, the concentrations of K in our modeling results are far below those of K observed in Na–HCO3-type springs, because of the precipitation of muscovite considered in the model, suggesting the partial disequilibrium state of the aquifer during the hydrolysis of K-feldspar under high PCO2 conditions. This result implies that to better predict long-term CO2–water–rock interactions in a geological storage site with abundant K-feldspar, the secondary K-bearing minerals should be carefully predicted, because a target aquifer can be far from chemical equilibrium during the storage period. This study shows that geochemical modeling can be effectively used to predict the hydrochemical changes of groundwater during long-term CO2–water–rock interactions and subsequent leakage toward surface in K-feldspar rich aquifer, although it should be included in a fully coupled computational approach between fluid flow, heat transfer and reactive mass transport processes in the future research.