Field-scale quantification of methane and nitrous oxide agricultural emissions using micrometeorological approaches

Abstract

Methane (CH₄) and nitrous oxide (N₂O) emissions from agricultural systems are significant contributors to global greenhouse gas (GHG) emissions, and accurate quantification of these emissions is essential for improving GHG inventories and evaluating mitigation strategies. This research employs the eddy covariance (EC) technique, artificial neural networks (ANN), and footprint analysis to address key challenges, such as spatial variability, data gaps, and footprint limitations, in measuring GHG emissions in dynamic agricultural environments. The objectives of this study were to: 1) assess the performance of the EC technique for measuring N₂O and CH₄ fluxes in agricultural systems; 2) develop ANN models to fill gaps in N₂O flux data; and 3) evaluate CH₄ emission rates from grazing systems using EC and footprint models. The first study focused on N₂O emissions measured using the EC technique in two contrasting agricultural environments: a cattle feedlot and a winter wheat field. In the cattle feedlot, average N₂O fluxes were low (0.0062 μ mol m⁻² s⁻¹) during winter, with emission spikes that were not consistently linked to precipitation or temperature changes due to frozen soil conditions. In contrast, the winter wheat field exhibited higher variability, with emission rates peaking at 0.03 μ mol m⁻² s⁻¹ during warm, moist periods following nitrogen fertilization and rainfall. Footprint analysis in the cattle feedlot revealed that up to 70% of the measured flux contributions came from outside the feedlot boundary, driven by manure runoff and off-site sources. In the winter wheat field, N₂O fluxes were more evenly distributed within the footprint and closely associated with fertilizer applications and subsequent precipitation events. These findings demonstrate the necessity of continuous high-frequency measurements to capture the full temporal and spatial variability of N₂O fluxes in agricultural systems. The second study applied artificial neural networks (ANN) to fill data gaps in N₂O flux measurements at winter wheat fields. Data gaps, caused by equipment malfunctions, low-turbulence conditions, and extreme weather events, can significantly affect the accuracy of long-term flux data. The ANN models were trained using environmental variables, such as soil moisture, temperature, and wind speed, and demonstrated strong performance, achieving a coefficient of determination (R² > 0.85) between predicted and observed fluxes. At the winter wheat field, the ANN model successfully reconstructed missing flux data due to cold weather and sensor failures and filled gaps during elevated N₂O emissions following fertilization. These results highlight the importance of using advanced gap-filling techniques, such as ANN, to maintain the integrity of long-term GHG datasets, particularly in complex agricultural systems. The third study focused on quantifying CH₄ emissions from grazing systems using the eddy covariance (EC) technique in combination with flux scaling based on footprint analysis. Controlled CH₄ release experiments were conducted to simulate emissions from a small herd of cattle, allowing for the assessment of the EC system's accuracy in estimating CH₄ fluxes. The results revealed that CH₄ fluxes were consistently underestimated, with the calculated emission rates clustering around 30% of the actual release rate. This underestimation was linked to the footprint model’s difficulty in fully capturing contributions from the source area, especially at lower fetch percentages. Increasing the fetch to 90% improved source coverage but introduced instability and errors. These findings emphasize the need for more accurate footprint models to account for varying wind conditions and terrain complexities in grazing environments. The integration of these studies underscores the importance of continuous high-frequency measurements in capturing the variability of N₂O and CH₄ emissions in agricultural systems. Due to uncontrollable factors such as equipment failures and adverse weather conditions, data gaps are inevitable, making gap-filling techniques like ANN crucial for ensuring data continuity. Footprint analysis further improves the accuracy of emission estimates by addressing spatial variability in emission sources, particularly in grazing systems. This research advances the understanding of GHG emissions from livestock and crop production systems, contributing to more accurate emission inventories and the development of effective mitigation strategies in agriculture.