Quantification of Volcanic CO2 Emissions Using the Eddy Covariance Method

Eddy covariance (EC) is a micrometeorological technique proposed as a method to measure passive volcanic CO₂ emissions from soil, vent, groundwater, and surface water sources. EC provides an automated, semi-continuous, and time and space-averaged CO₂ flux measurement. Also, the measurement's "intermediate" spatial scale (m²-km²) has the potential to bridge the gap between relatively small-scale ground-based measurements (e.g., using the accumulation chamber, AC, method) and relatively large-scale satellite-based observations. We deployed and tested an EC system during two studies in an area of diffuse volcanic CO₂ emissions on Mammoth Mountain, CA and near a bubbling spring in Soda Springs, ID. Half-hourly EC CO₂ fluxes were measured on Mammoth Mountain during September-October 2006 and ranged from 218 to 3500 g m^-2 d^-1. Maps of surface CO₂ flux were simulated based on AC measurements made repeatedly on a grid over a ten-day period. Large meteorologically driven variations in surface flux distributions and emission rates (16 to 52 t d^-1) were observed. Using source weight function modeling, we compared EC to AC measurements of CO₂ flux. Half-hour EC CO₂ fluxes were moderately correlated (R² = 0.42) with AC fluxes, whereas average-daily EC and AC fluxes were well correlated (R² = 0.70). We then made EC measurements of CO₂ flux on Mammoth Mountain during September-October 2010, which ranged from 85 to 1766 g m^-2 d^-1. Three AC soil CO₂ flux surveys during this time were used to simulate maps of soil CO₂ flux and estimate total emission rates. An inversion of measured EC CO₂ fluxes and corresponding modeled source weight functions was carried out and recovered 58 to 77% of the CO₂ emission rates estimated based on simulated AC soil CO₂ fluxes within a 0.01 km² area. Spatial distributions of modeled surface CO₂ fluxes based on EC and AC observations showed moderate to good correspondence (R² = 0.36 to 0.70). In September-October 2011, we deployed an EC system near a bubbling spring in Soda Springs, ID and measured CO₂ fluxes from -74 to 1147 g m^-2 d^-1. An inversion of measured EC CO₂ fluxes and modeled source weight functions mapped the surface CO₂ flux distribution within and quantified CO₂ emission rate (24.9 t d^-1) from a 0.05 km² area surrounding the spring. This emission rate was 79% of the dissolved CO₂ discharge from the spring estimated as the product of the measured concentration of dissolved inorganic carbon and spring outflow rate. Overall, the results of our investigations suggest that under appropriate terrain and atmospheric conditions, EC can serve as a valuable tool for semi-continuous monitoring and quantification of volcanic CO₂ emissions from moderate size land areas.