We have measured carbon dioxide fluxes of wetlands in Northern Wisconsin, USA, using the eddy covariance method. One of these sites, Lost Creek, has had continuous flux observations for six years. Lost Creek is a fen wetland with predominant ground cover of alder and willow shrubs. Measurements were taken using a 30 foot tower instrumented with a sonic anemometer and a high precision gas analyzer for measuring fluxes of carbon dioxide and water. The tower is also equipped to measure air temperature, humidity, and solar radiation. The equipment on the tower is supplemented with measurements of precipitation, water table height, soil temperature, moisture, and heat flux, tree stem temperatures, and leaf wetness. We analyzed a six year record of carbon dioxide data in conjunction with measurements of water table height, precipitation, and latent heat flux to characterize the wetland ecosystem response to changes in hydrology. A model was used to estimate ecosystem respiration and photosynthesis based on measurements.
Since water dynamics play such a crucial role in wetland ecology, we expect net fluxes of both CO2 and methane to respond to changes in water table height. Here we present initial results from a study of these connections. At Lost Creek, ecosystem respiration was found to be negatively correlated with water table height, with a correlation coefficient of approximately 0.6. Photosynthesis had a similar negative correlation with water table height. The respiration and photosynthesis responses offset, leading to a weak correlation of net ecosystem exchange with water table. We expect these relationships to hold at two other nearby open and forested wetlands where we have made shorter term measurements (two growing seasons) of CO2 flux, CH4 flux, and water table height. These results imply that complex interactions exist between wetland ecosystem carbon and water balances
To further investigate mechanisms and scale results to the region, we have also begun to incorporate wetland hydrology and biogeochemistry into an existing ecological model, TREES, which has previously been successfully used in the region to simulate transpiration. Observed carbon dioxide and latent heat flux data from wetlands are used to estimate parameter values and associated uncertainties using Bayesian methods. Initial results show promising performance from the model in capturing and predicting wetland flux dynamics.