Inferring methane fluxes at a larch forest using Lagrangian, Eulerian, and hybrid inverse models

Measuring methane (CH4) flux at upland forests is challenging due to high levels of heterogeneity in upscaling chamber measurements and the detection limits of currently available micrometeorological methods. We estimated CH4 fluxes in an upland forest from vertical concentration profiles using three different inverse multilayer models: the Lagrangian localized near field theory, Eulerian, and hybrid Lagrangian‐Eulerian models. The approach could estimate spatially representative fluxes, and use of higher gradients within canopies than above them could minimize uncertainties associated with sensor noises. Comparing fluxes by the models and measurements by the micrometeorological hyperbolic relaxed eddy accumulation and chamber methods, daytime fluxes were reasonably reproduced, but nighttime fluxes were overestimated most likely due to an underestimation of stable conditions and storage effects. The models and measurements show that the forest acted as a CH4 sink during the study period, and the soil acted as the dominant sink. The estimated sink increased with increasing soil temperatures and decreasing soil water content. The CH4 sink estimated during the study period were 1.5 ± 0.2 nmol m−2 s−1 by the micrometeorological method, 2.4 ± 0.5 nmol m−2 s−1 by chambers, 2.8 ± 1.1 nmol m−2 s−1 by the Lagrangian model, 2.7 ± 1.0 nmol m−2 s−1 by the Eulerian model, and 3.7 ± 2.8 nmol m−2 s−1 by the hybrid model. The performance of the Lagrangian and hybrid models increased when the CH4 sink/source was assumed to only exist in the soil. Key Points CH4 fluxes at a forest was estimated by three inverse models and CH4 profileEstimated flux was consistent with micrometeorological and chamber measurementsCH4 sink existed at the soil and varied with soil temperature and water content