Comparison of Four Approaches for Estimating Soil-Atmosphere Trace Gas Fluxes.

Accurate measurements of trace gas fluxes are critical in estimating the impact of agricultural management practices on greenhouse gas (GHG) emission.  Our objectives were to compare several approaches to estimate gas fluxes and to select a proper model for each gas. Gas samples were collected on 48 days from an agricultural field with three cropping systems using static-chamber technique, and N₂O, CO₂ and CH₄ concentrations were determined. A total of 1536 measurements were processed using four approaches: linear regression (LR), Hutchinson-Mosier (HM), Quadratic (QD), and Hutchinson-Mosier-Revised (HMR); and their performance was evaluated. The HMR is based on an implementation of the R statistical analysis package. The HM method requires three observations with equi-spaced time intervals, while HMR can be used with more than three observations.  The HMR provided the best approach to estimate N₂O and CO₂ fluxes in a single method and eliminated underestimation and overestimation of flux. The HM underestimated CO₂ flux compared to the HMR, and HMR and QD models were a best fit. The LR failed to estimate N₂O, CH₄ and CO₂ fluxes in 38, 47, and 73% of total measurements, respectively, compared to HMR. The HM failed to estimate CO₂, N₂O and CH₄ fluxes in 27, 74, and 75% of total measurements, respectively. The QD failed to estimate N₂O, CO₂, and CH₄ fluxes in 21, 27, and 45% of total measurements, respectively. Comparison of LR, QD and HM with the HMR procedure revealed that HMR was the best approach for N₂O and CO₂ flux estimation under a dryland cropping system while LR provided the best estimate for CH₄ flux estimation within the detection limit. Compared to LR, QD and HM, HMR is the most comprehensive, robust and flexible.