A New Approach to Address Soil Heterogeneity as a Source of Uncertainty in the Flux‐Gradient Method: <scp> CO ₂ </scp> Case Studies

ABSTRACT The flux‐gradient method (FGM) is an important tool to study soil‐atmosphere and subsurface gas fluxes. The simplicity of the approach can lead to an uncritical application. Typical uncertainty found in the input parameters is not considered in most cases. Their potential effect on the flux estimations might be negligible, but could also result in a relevant uncertainty or even bias. In this study, we investigated how measurement uncertainty and soil heterogeneity may affect the application of the FGM. We introduce a new analysis approach that allows to include (a) additional chamber measurements and (b) known parameter ranges/distribution of soil physical properties for model calibration and (c) to quantify the uncertainty in the flux estimate. The new Robust Calibrated Inverse FGM (RCI‐FGM) approach is an extension of the FGM and shared within the new R‐package ConFluxPro. In two soil CO 2 data studies, we demonstrate how soil heterogeneity affects gas flux estimations calculated with the FGM, and how RCI‐FGM helps to derive more robust flux estimates. In study 1, we found that scattering (due to measurement uncertainty and soil heterogeneity) found typically in the total porosity of soils can drastically change the vertical concentration profile of soil CO 2 . Assuming mean porosity in the FGM led to a significant bias in the estimated flux rates. The new RCI‐FGM approach successfully reduced this bias by incorporating reference flux measurements for calibration. In study 2, we applied the RCI‐FGM approach to a previously published dataset of forest soil CO 2 fluxes. RCI‐FGM improved the fit to reference chamber measurements and the plausibility of the vertical partitioning of the flux rates. The application of the RCI‐FGM approach in future studies was demonstrated for CO 2 fluxes but can be used for CH 4 and O 2 uptake in well‐aerated soils. For soils and processes dominated by hot spots or hot moments, such as N 2 O formation from denitrification, additional consideration may need to be taken. Our approach can help in future studies to address the uncertainty in the FGM method, improve the robustness of the estimated flux rates, and increase the comparability of studies.