Detecting CO2 leakage at offshore storage sites using the covariance between the partial pressure of CO2 and the saturation of dissolved oxygen in seawater

An option to detect CO2 leakage from deep geological storage under the seabed is monitoring the partial pressure of CO2 (pCO2) in the sea. One of the challenges of this technique is to differentiate pCO2 elevations caused by CO2 leakage from variations in the natural baseline (which varies due to seasonal and other factors). This paper compares a covariance threshold for anomalous pCO2 (a threshold based on pCO2 and dissolved oxygen), with annual and seasonal constant thresholds, using data observed in two hydrologically contrasting areas in Osaka Bay, Japan. In one of these locations the water column is typically stratified throughout the year, whereas at the other it is vertically mixed. The covariance threshold often but not consistently outperforms the constant thresholds, but is effective in all the cases analyzed in the present study. In any season in the mixed area, and in winter in the stratified area, the possibility of false-negatives in the seasonal threshold method is smaller than or comparable with that in the covariance threshold one. In areas with large natural pCO2 variations, such as during summer in the stratified area, the covariance threshold method is efficient in detecting CO2 leakage whereas the seasonal threshold method is of little use. With a good understanding of local conditions, monitoring pCO2 with the covariance threshold method is a potentially useful but not sufficient option for leakage detection in the marine environment.