Modification of a Wavelet-Based Method for Detecting Ebullitive Methane Fluxes in Eddy-Covariance Observations: Application at Two Rice Fields

Ebullition, the release of gas bubbles, is an important pathway of methane emission in many ecosystems, yet its high spatio–temporal variability makes it challenging to quantify. In this work, a methane-flux partitioning method based on scalar similarity in the wavelet domain is applied to eddy-covariance data collected at two flooded rice fields. Inspection of initial results indicates that several modifications are needed for robust ebullition detection. With these modifications, our objectives are to compare the original and modified methods, conduct a sensitivity analysis of the program’s empirical parameters, characterize the importance of ebullition in rice across growth stages, and identify the primary drivers of ebullition. The modified method’s ebullitive fluxes are significantly lower and show lower variance than those from the original method. Furthermore, the two methods produce distinct patterns of diel variation. While partitioning estimates show non-trivial sensitivity to the program parameters, this sensitivity is lower in magnitude than the random error in the ebullitive flux estimates. Ebullitive fluxes make up 9% of the total flux on average, with ebullition increasing in importance as plants develop. Ebullitive fluxes are best predicted by wind speed (negative effect), ecosystem respiration (positive effect), and sensible heat flux (positive effect), suggesting an indirect effect of plant-mediated transport, a link with temperature and methane production, and a potential effect of water column turnover, respectively. In addition to validating the method with independent ebullition observations, we recommend its application at more natural and managed wetlands to improve understanding of this highly variable transport pathway.