The study of soil–atmosphere gas exchange has become more prominent over the past couple of decades. Objectives for these studies are wide-ranging, for example, the study of volcanic degassing (Carn et al., 2016; Cardellini et al., 2017), quantification of carbon budgets (Houghton and Nassikas, 2017; Le Quéré et al., 2018), greenhouse gas (GHG) emission studies (Oertel et al., 2016), and identifying potential leakage from enhanced oil recovery (EOR) and carbon capture and storage (CCS) sites (Korre et al., 2011; Beaubien et al., 2013; Jones et al., 2014). Soil gas emissions are directly measured at points or spots using chamber techniques (Pumpanen et al., 2004) or over restricted areas through micrometeorological methods (Dugas, 1993). At regional and national scales, airborne and space measurements are used to derive soil gas emissions using empirical and process-oriented models for post-processing. These regional-scale methods lack detail required for field-scale studies (101 to 103 m2) and may be prohibitively expensive (Oertel et al., 2016). Feitz et al. (2018) provide a comparison of many of these techniques under a controlled gas release.

Closed-loop flux chamber-based analyses utilize an open-bottomed chamber with a known footprint and volume placed on the soil surface, allowing gases emitted by the soil to accumulate within the chamber headspace (Rolston, 1986). From analysis of the gas mixing ratios within the chamber over time, the flux of gas from the soil can be derived for that small spot of the land surface. In contrast, an open-loop technique passes air through the sample chamber at a known flow rate, until a steady-state concentration is observed, from which a flux rate is derived (Denmead, 2008). Both techniques require measurements at a large number of points to estimate field-scale fluxes via interpolation, with the caveats that sample density is sufficient to represent site spatial variability and that flux is static with respect to time during the measurement period (Gao et al., 1998). The use of chambers to represent soil gas fluxes at field scale is often highly time-intensive and prone to interpolation-related uncertainty (Elío et al., 2016), particularly for sites with heterogeneous soils or geology.

Micrometeorological methods use eddy covariance (EC) techniques to derive soil gas flux. A three-dimensional sonic anemometer is coupled to a gas analyser attached to a tower or mast, allowing measurements that incorporate areas up to several square kilometres under the right atmospheric and terrain conditions (Myklebust et al., 2008). Continuous EC measurements over a period of time (usually days to weeks) allow soil gas fluxes for a particular parcel of land (the EC footprint) to be derived from absolute gas concentrations, temperature, and vertical and horizontal wind flows (see for example, Eugster and Merbold, 2015). The EC footprint location and size are calculated through post-processing of the high-resolution data, averaged over longer time intervals (Aubinet et al., 2012). EC methods require a fully turbulent flux, where the majority of vertical movement is driven by eddies, and uniform, homogeneous terrain, where air density fluctuations and convergence/divergence are negligible (Lee et al., 2004). Soil gas flux from EC is derived by integrating the net fluxes upwind from the measurement point (Eugster and Merbold, 2015). A key component of the EC method is calculating (a posteriori) the pathway from the instrument sensors to the soil surface under turbulent conditions, which leads to multiple assumptions (see Baldocchi and Meyers, 1998). Tower- or tripod-based EC methods are difficult to utilize for consistent identification of soil gas flux at any particular location, as they are reliant on wind direction, surface roughness, and atmospheric conditions to determine the location of their footprint. Roving or mobile EC towers can effectively enlarge the EC footprint to cover any particular location; however these techniques take days to weeks of measurement to provide sufficient coverage at the field scale (Eugster et al., 1997; Billesbach et al., 2004).

Unmanned aerial vehicles (UAVs) could potentially be used to capture field-scale flux in the future, but currently they have limited flight times (<30min) and weight restrictions that limit sensor options to small, low-power devices (see for example, Danilov et al., 2015; Hass et al., 2014). These low-power sensors do not currently have the sensitivity to observe small flux anomalies at flight height and speeds of fixed-wing UAVs, and slower copter-style UAVs generate too much downdraft for an accurate measurement (Li et al., 2020).

There is currently a lack of practical, fast, inexpensive methods for quantifying soil gas flux at field scales, which are highly relevant to leakage and degassing studies. The objective of this paper is present a new “open-field” method that uses aspects of chamber and micrometeorological methods combined with a mobile platform and GPS to rapidly derive soil gas fluxes at the field scale. We assess this method against traditional chamber techniques for field locations within the UK and Italy and discuss the explicit and implicit assumptions inherent in the presented technique.

Development of a new field-scale soil CO2 flux quantification method was focused on creating a mobile tool that could easily and quickly make measurements around a field site without the need for stopping at individual locations, and that was valid on sloping or heterogeneous terrain. Here we describe the theoretical aspects, assumptions made, components used to undertake the measurements, and post-processing requirements.

Figure 2 shows CO2 flux data (gm-2d-1, note the difference in temporal units compared to the flux equation presented in the methods section) from a single field at the Ailano site, with fluxes from the chamber method plotted on top of the interpolated field-scale flux distribution. A visual comparison shows that the open field-scale method produces values that are of a similar order of magnitude to those obtained by chamber methods, although there are clear differences at individual observation locations. CO2 flux data for each individual survey point at both field locations are plotted in Fig. 3.

To quantitatively compare open field-scale and closed-loop chamber methods at the larger scale, we have de-localized the datasets and derived a set of summary statistics. The difference in mean CO2 flux values between the techniques is 0.5 gm-2d-1 for Ailano and 2.4 gm-2d-1 for Sutton Bonington. Figure 4 shows box-and-whisker plots representing the rest of the summary statistics for both sites. The median for both techniques (50 % percentile) is highly comparable for the two sites, while the open field-scale method exhibits less range between the 25 % and 75 % percentiles. Regression analysis (Fig. 5) shows the deviation from a 1:1 relationship between the point and mobile flux techniques. The coefficient of determination (r2) between the techniques is 0.29 for Ailano and 0.08 for Sutton Bonington.

The results show that the average (mean and median) CO2 flux obtained using the chamber technique and those derived from the open field-scale method presented in this study are highly comparable at field-study scales. The range (absolute and between quartiles) of CO2 flux is smaller in the latter dataset, which, as discussed in the assumptions section, is likely due to atmospheric dilution.

At individual measurement locations there is the potential for correlation between the chamber and open-field-derived techniques, as shown at the Ailano site. However, this was not apparent in the data collected for Sutton Bonington. This could be due to atmospheric conditions or that the observed flux values were much smaller at the UK site. We assessed wind speed and direction against the difference in flux between the two techniques, with low regression values indicating no relationships between the wind variables and differences in flux (see Supplement).

The present work describes the theoretical basis and preliminary test results of a new method for rapidly estimating CO2 flux from the ground surface over large areas, opening the door for future development and improvement of this hybrid approach. The developed “open-field” method, which combines aspects of open-chamber and micrometeorological methods on a mobile platform, is less fieldwork intensive than traditional chamber techniques and cheaper than those derived from airborne or space surveys. Due to several assumptions, the most accurate results are expected under stable atmospheric conditions, with little horizontal wind flow. When derived soil gas fluxes are averaged at the field scale, they are highly comparable to results obtained using traditional chamber techniques. As expected, atmospheric dilution leads to a reduced range of flux values under the open field-scale method. Under ideal atmospheric conditions it may be possible to use the new method to derive soil gas flux at an individual point; however this requires further investigation. The presented method of deriving soil–atmosphere gas exchange at the field scale could be useful for a number of applications including leakage, degassing, and greenhouse gas emission studies. The results presented for the open-field flux method are limited in scope, and it is recognized that further research is required to assess robustness under different environmental and meteorological conditions.