This study examined a mountainous area with two hydrochemically distinct CO-rich springs to understand the origin, flow, and leakage of CO, which may provide implications for precise monitoring of CO leakage in geological carbon storage (GCS) sites. The carbon isotopic compositions of dissolved inorganic carbon (DIC) in CO-rich water (δC) and those of soil CO (δC) indicated a deep-seated CO supply to the near-surface environment in the study area. The hydrochemical difference (e.g. pH, total dissolved solids) for the two CO-rich springs separated by 7 m, despite similar δC and partial pressure of CO, was considered as the result of different evolution of shallow groundwater affected by deep-seated CO preferentially rising along fracture zones. Electrical resistivity tomography also suggested flow through fracture zones beneath the CO-rich springs, showing low resistivity compared to other surveyed zones. However, soil CO efflux was low compared to that in other natural CO emission sites, and in particular it was noticeably low near the CO-rich springs, whereas δC was high close the CO-rich springs. The dissolution of CO in the near-surface water body seemed to decrease the deep-seated CO leakage through the soil layer, while δC imprinted the source. End-member mixing analysis was performed to assess the contribution of deep-seated CO to the low soil CO efflux by assuming that atmospheric CO and soil CO (by respiration) as well as deep-seated CO contribute to the soil CO efflux. For each end-member, characteristic δC and CO concentrations were defined, and then their apportionment to soil CO efflux was estimated. The resultant proportion of deep-seated CO was up to 8.8%. Unlike the spatial distribution of high soil CO efflux, high proportions exceeding 3% were found around the CO-rich springs along the east-west valley. The study results indicate that soil CO efflux measurement should be combined with carbon isotopic analysis in GCS sites for CO leakage detection because CO dissolution in the underground water body may blur leakage detection on the surface. The implication of this study is the need to quantitatively assess the contribution of deep-seated CO using the soil CO concentration, soil CO efflux, and δC at each measurement site.Copyright © 2020 Elsevier Ltd. All rights reserved.