COS 172-8 - Do high-frequency sensors help reduce uncertainty in greenhouse gas emissions from wetlands?

Friday, August 11, 2017: 10:30 AM
D132, Oregon Convention Center
Ashley R. Smyth, Kansas Biological Survey, University of Kansas, Terry D. Loecke, 2101 Constant Ave., University of Kansas, Lawerence, KS and Amy J. Burgin, Kansas Biological Survey, University of Kansas, Lawrence, KS
Background/Question/Methods

Global greenhouse gas (GHG) emissions from wetland ecosystem are highly variable. Accurate accounting of fluxes from wetlands is challenging because these systems experience rapid periods of drying and wetting. Although methane (CH4) and nitrous oxide (N2O) fluxes have been measured or simulated at different spatial and temporal scales, long term GHG measurements and accompanying environmental data are needed to reduce uncertainty and identify key drivers of gas emissions. To evaluate fluxes and drivers of CH4 and N2O emissions we used a soil-monitoring network located in a restored wetland near Dayton, Ohio. By coupling weekly GHG flux measurements with high-frequency soil sensor data for oxygen (O2), water-filled pore-space and temperature, we investigated whether sensor data could improve our understanding of greenhouse gas emissions. Using data from 24 sites collected between 2012-2014 we developed rolling-window models, based on pre-defined time intervals, and instant models, based on sensor readings recorded at the same time as GHG flux measurements, to predict greenhouse gas emissions. In our system, soil temperature, moisture and O2were not correlated and considered as additive predictors in alternative models.

Results/Conclusions

Models which included continuously recorded sensor data improved predictions of GHG emission compared to discrete measurements (ΔAIC=852.30 for CH4 and 680.58 for N2O). CH4 emissions were driven primarily by soil O2 concentration at the time of the flux as well as the variance in O2 over the previous 24-hours (AIC weight=0.72). The highest CH4 fluxes occurred when soil O2 was low following variable soil O2 the day prior. This indicates that short-term fluctuations in soil O2 lead to increased CH4 emissions. N2O emissions were best predicted by soil temperature, however water-filled pore-space and soil O2 preformed equally as well. While individual sensors improved estimates of GHG emissions compared to models without sensor data, combining data from soil temperature, O2 and water-filled pore space explained more of the variance in GHG fluxes. Our results indicate that long-term continuous data can improve our understanding of wetland ecosystem function and aid in the development of effective management strategies to mitigate climate impacts.