Using high-frequency soil oxygen sensors to predict greenhouse gas emissions from wetlands

Abstract

Understanding the mechanisms that lead to greenhouse gas (GHG) emissions requires knowledge of short-term fluctuations in drivers of fluxes. To better understand how variation and lags in abiotic conditions affect soil GHG fluxes, we coupled weekly methane (CH4) and nitrous oxide (N2O) flux measurements with time-series soil sensor data for oxygen (O2), moisture and temperature. Including time-series data improved models for soil CH4 and N2O fluxes compared to models which did not contain time series data. Fluctuations in soil O2 drove CH4 fluxes, where rapid changes in redox conditions led to high fluxes. N2O fluxes occurred when soils were warm and dry. Soil O2 was the best predictor and thus sensor for understanding CH4 fluxes, whereas soil moisture best predicted N2O fluxes. Combining data from multiple sensors improved models for both gases, underscoring the relative importance of interactions among drivers. Overall, our top models explained 15% and 30% of the variance in N2O and CH4 fluxes, respectively. Fluxes predicted using linear interpolation between measured fluxes were lower than time-series model predictions for both N2O and CH4 fluxes. This suggests linear interpolation may fail to capture “hot moments” or episodic events which lead to high fluxes. Long-term, continuous data from sensors, which accounts for short-term variation in abiotic drivers, may improve our ability to predict the timing and intensity of GHG emissions from wetland soils.