[H-SC04] CCUS (Carbon Dioxide Capture, Utilization, and Storage) for Climate Mitigation

[HSC04-08] Is observing CO₂ concentration in seawater effective to detect CO₂ leakage?

*Keisuke Uchimoto^1,2, Yuji Watanabe^1,2, Humio Mitsudera³, Ziqiu Xue^1,2 (1.Geological Carbon dioxide Storage Technology Research Association, 2.Research Institute of Innovative Technology for the Earth, 3.Institute of Low Temperature Science, Hokkaido University)

Keywords:offshore CO2 geological storage, monitoring for leakage, base rate fallacy, ocean simulation, pCO2, false positive

Observing a CO₂ concentration index in seawater such as pCO₂ and pH (hereafter pCO₂ is used) is thought to be a method to detect CO₂ leakage at offshore CO₂ storage sites. This is based on the thought that observing pCO₂ tells us sign of CO₂ leakage since pCO₂ in seawater increases in the event of CO₂ leakage. CO₂ release experiments conducted in Europe actually showed that pCO₂ in seawater near the release point clearly differed from that before the release or at reference points unaffected by the release, and, thus, implied that observing pCO₂ would be useful in the monitoring. We should, however, notice that showing anomalous pCO₂ at a known release point greatly differs from detecting anomalous pCO₂ due to leakage, especially when it is not certain whether CO₂ is even leaking. In the present study, we reconsidered whether observing pCO₂ in seawater is an effective method to detect CO₂ leakage. First, we investigated the area where we may find pCO₂ exceeding the threshold (anomalous pCO₂) due to leakage, using a simulation of passive tracer dispersion. We assumed that the increase in pCO₂ due to leakage (ΔpCO₂) of 90 μatm exceeded the threshold with a probability of 50 % and that CO₂ leakage was potentially detectable where pCO₂ is larger than 90 μatm for over half the period. The results showed that such an area was only within a 1 km by 0.5 km rectangular around the leakage point even when the leakage rate is 10,000 tonnes/y in summer when ΔpCO₂ can be the highest. Given that even the shortest distance between observation points in the Tomakomai Project is about 1 km, it is not feasible to observe pCO₂ in such a dense way, and, thus, detecting leakage by observing pCO₂ would be unlikely. Second, we considered false positives. When the threshold is the upper limit of the 95 % prediction interval like the Tomakomai Project, the threshold would be exceeded 1 out of every 40 times without leakage. If the upper limit of the 99 % prediction interval is used, the threshold would be exceeded only 1 out of 200. This may make us intuitively consider that pCO₂ exceeding the threshold probably indicates CO₂ leakage because pCO₂ rarely exceeds the threshold without leakage. This intuitive estimation is, however, false, known as the base rate fallacy. When the storage site is properly selected and managed, CO₂ leakage is thought to be unlikely to occur. If the probability of CO₂ leakage, which we define here as the probability that not only CO₂ leaks but also an observation point is within the 1 km by 0.5 km rectangular, is assumed to be 0.01 %, the chance of true CO₂ leakage is only 0.5 % when pCO₂ exceeds the threshold. The remaining 99.5 % is attributed to the natural variability. This suggests that most of data exceeding the threshold would be false positives. The chance of CO₂ leakage is, however, 85 %, when the probability of CO₂ leakage is assumed to be 10 %. These results imply that observing pCO₂ is useless for leakage detection when there are no sign of CO₂ leakage but that it is useful when CO₂ is likely to leak; that is when the CO₂ arrives at the shallow formations after its leakage is detected in the deep-focused monitoring. To conclude, we consider that an appropriate monitoring to detect CO₂ leakage is the following way. The reservoir and deep formations should usually be monitored. If leakage is detected, the leaked CO₂ should be traced using a high resolution monitoring system such as the P-Cable. It is after where and when CO₂ would leak out can be able to be estimated that marine monitoring such as CO₂ bubble search and/or pCO₂ observation should be conducted.
This paper is based on results obtained from a project (JPNP18006) commissioned by the New Energy and Industrial Technology Development Organization(NEDO) and the Ministry of Economy,Trade and Industry (METI) of Japan. We used the ocean model, kinaco, developed by Dr. Yoshimasa Matsumura at Atmosphere and Ocean Research Institute, The University of Tokyo.