Geological storage of CO2 is a leading strategy for large-scale greenhouse gas emission mitigation. Monitoring and verification is important for assuring that CO2 storage poses minimal risk to people's health and the environment, and that it is effective at reducing anthropogenic CO2 emissions. Eddy Covariance (EC) has been proposed as a long-term monitoring solution for geological storage projects and is considered suitable for monitoring areas 1000 - 100,000 m2 in size. Eddy Covariance is a key micrometeorological technique which has traditionally been used for assessing ecosystem exchange of CO2 in a variety of natural and agricultural settings. It measures the vertical transfer of scalar variables such as CO2 via eddies from upwind of the instrumentation, and correlates the measured CO2 flux to the upwind source area based on several key assumptions. These assumptions include that the upwind source area is homogeneous, flat and uniform, which in turn requires that horizontal gradients in CO2 concentration are zero and that horizontal and vertical gradients in the covariance of CO2 concentration and orthogonal wind directions are zero. Work undertaken at the GA-CO2CRC Gininnderra controlled release facility, where CO2 is released from the shallow subsurface (at 2 m depth), suggests that CO2 leakage in the near subsurface will follow paths of least resistance up to the surface. Similar observations have been observed at the ZERT facility in Montana and CO2 Field Lab in Norway. This leads to CO2 leaks having localised, patchy surface expression, rather than a diffuse wide-scale leak which one typically expects (Lewicki et al. 2010). The implication of this is that the source area for a leak is highly inhomogeneous, meaning the magnitudes of CO2 flux values measured using EC are grossly unreliable. These limitations were discussed in Leuning et al.'s (2008) review on CCS atmospheric monitoring technologies yet are not addressed in much of the recent EC leak quantification literature. This presentation will present findings from the first subsurface release at the CO2CRC facility in Canberra (March - May 2012), where EC data was analysed for application in leak detection and quantification. The CO2 release rate was 144 kg/d. Eddy Covariance was successfully used to detect the leak by comparing CO2 fluxes in the direction of the leak to baseline wind sectors. Median CO2 fluxes in the leak direction were 9.1 µmol/m2/s, while the median background flux was 1.0 µmol/m2/s. Separate measurements taken using a soil flux meter found that the daytime background soil flux had a median flux of 1.8 µmol/m2/s but the peak soil flux over a leak was 1100 µmol/m2/s. Quantification and spatially locating the leak were attempted, but due to the problem of source area inhomogeneity, no substantive progress could be made. How an inhomogeneous source area contributes to 'lost' CO2 from the system, through advection and diffusion, will be discussed, coupled with suggestions for how these parameters can be evaluated in future experimental design. Leuning R., Etheridge D., Luhar A., and Dunse B., 2008. Atmospheric monitoring and verification technologies for CO2 sequestration. International Journal of Greenhouse Gas Control, 2(3), 401-414. Lewicki J. L., Hilley G. E., Dobeck L., and Spangler L., 2010. Dynamics of CO2 fluxes and concentrations during a shallow subsurface CO2 release. Environmental Earth Sciences, 60(2), 285-297. Presented at the 2014 Australian Earth Sciences Convention (AESC)

Two shallow sub-surface CO2 controlled release experiments were conducted at the Ginninderra test site during 2012. The theme of the first experiment was CO2 detection in the soil and surface emissions quantification. The theme for the second experiment was investigating sub-surface migration and broad scale detection technologies. Our objective overall is to design cheaper monitoring technologies to evaluate leakage and environmental impact in the shallow sub-surface. Over 10 different monitoring techniques were evaluated at the site against a known CO2 release. These included soil gas, soil CO2 flux, soil analysis, eddy covariance, atmospheric tomography, noble gas tracers, ground penetrating radar, electromagnetic surveys, airborne hyperspectral, in-field phenotyping (thermal, hyperspectral and 3D imaging), and microbial soil genomics. Technique highlights and an assessment of the implications for large scale storage are presented in the following corresponding talks. Presented at the 2013 CO2CRC Research Symposium

Understanding the near surface migration patterns and rates of efflux of CO₂ is important for developing effective monitoring and verification programs for the geological storage of CO₂. Soil flux surveys are a well-established technique for characterising surface CO₂ emission sources from controlled release sites, CO₂storage sites or natural CO₂seeps. The performance of four interpolation methods; arithmetic mean (AM), two minimum variance unbiased estimators (MVUE), and a newly developed geostatistical cubic surface were evaluated using 21 soil flux surveys conducted over two controlled release experiments in 2012 and 2013, at the Ginninderra controlled release facility, Australia. Data was binned to approximate a regular sampling grid for improved performance of the whole-of-field AM and MVUE averaging techniques. The AM and MVUE methods were highly sensitive to deviations in the statistical distribution of the data, and performed inconsistently across the two experiments. These two methods proved ill-suited for application to CO₂ leak quantification due to their inflexible sampling and distribution requirements. The cubic technique provided the best net emission estimates across both experiments, and when applied at different bin sizes, estimating the true release rate to within 20% for the 2012 experiment and 45% below the release rate for the 2013 experiment. The cubic method is well-suited for CO₂ leak quantification because it is not limited by assumptions of the data’s spatial or statistical distribution. Net H₂O emissions of 29 kg/d were observed coincident with the high CO₂ flux zones in the field. The interpolation methods were applied with similar results on soil flux surveys taken from a natural seepage site in Qinghai, China. Gravity currents appear to describe the observed soil flux and soil gas behavior at Ginninderra, i.e. the observed lateral migration of CO₂in the subsurface. Subsurface migration was also strongly influenced by the relative depth of the groundwater. Thus the low water table and greater vadose zone in the 2013 experiment is suspected to facilitate greater lateral CO₂ migration and explain the poor closure of the CO₂ balance. <b>Citation:</b> I.F. Schroder, P. Wilson, A.F. Feitz, J. Ennis-King, <i>Evaluating the Performance of Soil Flux Surveys and Inversion Methods for Quantification of CO2 Leakage</i>, Energy Procedia, Volume 114, 2017, Pages 3679-3694, ISSN 1876-6102, https://doi.org/10.1016/j.egypro.2017.03.1499.

Eddy Covariance (EC) has been proposed as a surface monitoring solution for long-term deployment at CCS sites. However, its suitability when applied to a highly inhomogeneous source area- as would be the case for a small-scale CO2 surface leak- has been poorly established. For this reason, EC has been implemented for two controlled CO2 releases conducted at the Ginninderra controlled release facility, with the aim of determining the technique's suitability for the location, detection and quantification of a small magnitude CO2 leak (144 kg/d). By comparing results from the two release experiments, this poster highlights the variable success of using EC for detection, and how this may depend on changing experimental and climatic variables such as leak location, tower height and depth to groundwater. The detection significance of grouped EC measurements will be established through statistical analysis using Cramer-Von Mises tests. In addition, the application of two EC towers concurrently for leak detection and location will be explored, with a second tower deployed for the latter portion of the 2013 release experiment. Quantification of the leak using EC was attempted, but due to the problems in the fundamental assumptions of the technique, no substantive progress could be made. This will be explained with respect to the 'lost' CO2 from the system in part due to advection and diffusion. Presented at the 2014 CO2CRC Research Symposium