Evaluation of new flux attribution methods for mapping N₂O emissions at the landscape scale

Highlights

• N₂O fluxes measured at different scales agreed in magnitude and temporal dynamics.

• New flux attribution methods were developed to map N2O fluxes over the landscape.

• Considering only land use as a control factor led to the best flux attribution.

• N₂O fluxes were driven by soil humidity and nitrate content.

• Catch-crop pea and corn fields emitted more N₂O than wheat, rapeseed and forest.

Abstract

The spatial variability of soil nitrous oxide (N₂O) fluxes is large − regardless of the study scale − resulting in very large uncertainties in soil N₂O emission assessments. The objectives of this study were to assess the N₂O flux at the landscape scale by coupling the results of measurements performed at different scales and to propose a method for obtaining emission maps based on these results. During a 2-month campaign (mid-March to mid-May 2015), N₂O fluxes were measured in a small cropland area (∼km²) (i) continuously at the plot scale using automatic chambers in a wheat field, (ii) punctually on a group of 16 plots including different types of soils and crops using a mobile chamber (fast-box), and (iii) continuously at the landscape scale by eddy covariance using a 15-m height mast. The soil properties were measured at all sites to provide a better understanding of the factors controlling the variability of the N₂O flux. To map the N₂O emissions of the entire area, two flux attribution methods were evaluated which allowed estimating the N₂O flux of each field during the whole period. These methods used a footprint model in combination with fast-box measurements over each crop type to determine the contribution of each field to the flux measured at the eddy covariance mast. Two footprint models were compared (the FIDES, and the Kormann and Meixner models) and two hypotheses on the dependency of N₂O emissions on crop distribution and soil nitrate contents were tested. Automatic chambers were used to evaluate the attribution methods. The N₂O fluxes measured by the different methods showed good agreement in magnitude and temporal dynamics, especially when the automatic chambers were in the eddy covariance mast footprint. Overall, the mean measured N₂O emission was 53 ± 6 μg NN₂O m⁻² h⁻¹ for the automatic chambers, 45 ± 7 NN₂O m⁻² h⁻¹ for the eddy covariance system and 37 ± 9 NN₂O m⁻² h⁻¹ for the fast-box, for periods when both automatic measurement systems were functioning. The N₂O fluxes measured by the automatic chambers and the fast-box were positively correlated with soil humidity (p < 0.01), water-filled pore space (p < 0.01) and nitrate soil content (p < 0.05). Catch crop-pea and catch crop-corn fields emitted more N₂O than wheat and rapeseed fields, and much more than forests. Over the whole area during the 2-month experimental period, the N₂O flux varied from 0.18 to 0.44 kg NN₂O ha⁻¹ month⁻¹ depending on the attribution method and footprint model. The simplest flux attribution method, taking only land use into account, showed very good agreement with the field measurements provided by the automated chambers (10%–13% difference on the mean flux). Our study demonstrates the potential of flux attribution methods for catching spatial variability of soil N₂O emission at the landscape scale and reducing uncertainties in its evaluation.

Introduction

N₂O has been the subject of concern due to its impact on global warming and ozone layer depletion (UNEP, 2013). There are several N₂O sources, but agriculture is by far the main anthropogenic N₂O emission source. Indeed, 77% of N₂O emissions due to anthropogenic activity are estimated to come from mineral and organic fertilizer application to agricultural land and other agricultural sources (UNEP, 2013). Monitoring N₂O emissions from agricultural fields is a key step to better constraining N₂O sources and determining the underlying processes driving emissions in order to develop mitigation recommendations. N₂O is produced at the microscale mainly by microbial processes, but the effects of its increased atmospheric concentration are visible at the global scale. N₂O emissions can be studied at the aggregate, field, landscape, regional, national and global scales. Regardless of the scale, the very high spatial and temporal variability of N₂O emissions from soil make its quantification difficult (Henault et al., 2012). Furthermore, N₂O emissions depend on numerous factors such as crop type, fertilizer type, and N application rate, as well as soil properties, such as soil organic carbon content, humidity, pH, and texture (Stehfest and Bouwman, 2006).

Several measurement methods have been developed to monitor N₂O emissions, with increasing accuracy, providing flux estimates at different scales. Chambers, which provide a gas-tight enclosure of a certain soil surface area while measuring the changes in gas concentrations in the chamber headspace, provide flux estimates at the plot scale. They can be used to assess (i) the spatial variability of N₂O fluxes if they are mobile (Flechard et al., 2007, Grossel et al., 2014) and (ii) the temporal variability if they are static, i.e., remain at a fixed position (Laville et al., 1997, Henault et al., 1998). High-frequency measurements can be obtained using automatic chambers (Christensen et al., 1996, ButterbachBahl et al., 1997, Neftel et al., 2010). The use of a micrometeorological mast/tower − to estimate the fluxes from meteorological conditions and gas concentrations − based either on eddy covariance (EC) (Christensen et al., 1996, Laville et al., 1999, Flechard et al., 2007, Molodovskaya et al., 2011), relaxed eddy accumulation (Baker, 2000, McInnes and Heilman, 2005), or the flux gradient method (WagnerRiddle et al., 1997, Desjardins et al., 2010, Zhang et al., 2015) provides information on the temporal variability over a wide spatial area, the so-called footprint, which is defined mainly by the height of the mast (plot, landscape and regional scale). Using aircraft (Pattey et al., 2006, Desjardins et al., 2010) to measure N₂O in the atmosphere provides information at an even larger scale (regional). These methods have been used to measure the biosphere-atmosphere exchange of trace gases for different ecosystems, different seasons and for different durations ranging from 4 days to 4 years (e.g. (Molodovskaya et al., 2011) (Desjardins et al., 2010).

Each of the many methods used to measure N₂O emissions from soils has specific advantages and limitations (Denmead, 2008). The fluxes obtained by the chamber method are generally associated with very large uncertainties because (i) the very small surface area covered by the chamber is not representative of the ecosystem and (ii) the gas-tight closure of the chamber affects natural gas diffusion gradients across the soil-atmosphere-interface (Davidson et al., 2002), which is a non-linear phenomenon and should be considered when calculating the flux (Pihlatie et al., 2013). The fluxes obtained by micrometeorological methods are dependent on the wind conditions during the measurement period, therefore, they do not represent a static area over time. Moreover, micrometeorological flux measurements are most suitable for flat terrain and provide only an integrated flux over a varying footprint area. Their installation, maintenance and data evaluation require a very high technical level and are expensive in terms of instrumentation and scientific man-power (Henault et al., 2012). Schematically, micrometeorological methods are the most appropriate for estimating gas fluxes in situations representative of ecosystems while chamber methods are the most appropriate for comparing emissions related to different agricultural practices.

For micrometeorological methods, the footprint area depends on the measurement height, so higher masts can be used to survey entire landscapes. Nevertheless, these methods rely on assumptions of horizontal homogeneity and non-advective conditions; data collected in complex terrain are difficult to analyze and the choices of sites are biased towards flat and homogeneous areas (Novick et al., 2014). While studies have addressed these problems for ecosystem CO₂ fluxes (Gockede et al., 2004), this has so far not be done for N₂O fluxes. N₂O fluxes are more difficult to measure due to atmospheric concentrations in the ppb range and because the pulses of N₂O fluxes are often sporadic. Additionally, high spatial variability needs to be considered. Fertilized croplands have been shown to be important sources of N₂O emissions, but crop type, fertilization and field management across a landscape is mosaic, challenging the assumption of the homogeneity of EC measurements.

Combining EC measurements with chamber measurements has been proven useful as it provides additional information on spatial variability as well as the temporal variability of fluxes for defined plots (Eugster and Merbold, 2015). Two types of approaches are currently available to link micrometeorological measurements with chamber measurements. The first is the bottom-up approach, which involves: (i) direct measurements at the soil surface and (ii) a method for extrapolating local results to larger scales of time and space, using basic extrapolation or ecosystem models such as Landscape-DNDC (Haas et al., 2013) or CERES-NOE (Gabrielle et al., 2006)). The second is the top-down approach, which involves (i) integrative micrometeorological measurements and (ii) atmospheric transport modeling. At the landscape scale, the Kormann and Meixner model (K&M) (Kormann and Meixner, 2001) and the FIDES model (Loubet et al., 2001, Loubet et al., 2009, Loubet et al., 2010) calculate the contribution of fields to the total footprint involved in N₂O flux detection dependent on meteorological conditions. Although relevant, this information is insufficient for describing the spatial variability of N₂O emissions. Understanding this spatial variability (especially at the landscape scale) remains a key concern for mitigating soil N₂O emissions.

In this study, different methods were used in parallel to assess N₂O emissions from the plot to landscape scale for a cropland area; a mobile chamber and static automatic chambers were used to measure N₂O fluxes at the plot scale, while an EC system was used to measure fluxes at the landscape scale. Several scientific issues were raised: (1) Are N₂O fluxes measured using several methods and covering several scales of the landscape comparable? (2) Can the spatial variability of the N₂O flux be evaluated over a 1-km radius landscape based on a single temporal dynamic flux measurement and several spatially spread chambers? (3) What is the integrated N₂O flux and the spatial variability of this flux over a 1-km radius landscape occupied by a mix of croplands and forests? Finally, (4) which method would be the best, and how can its validity and uncertainty be estimated?