Modeling C and N Dynamics in a Forested Wetland Watershed on Southeast Atlantic Coastal Plain, USA

Modeling dynamics of carbon (C) and nitrogen (N) in wetland watersheds requires spatial information of hydrology regulating the biogeochemical cycle in wetland ecosystems. To this end, we linked MIKE SHE, which is a distributed hydrologic model, to Forest-DNDC, which is a process-based biogeochemical model. To test the linkage, a 160 ha forested wetland watershed on the Atlantic Coastal Plain in South Carolina was divided into 665 cells with a 50x50 m grids; and water table, stream flow and soil respiration dynamics were observed in the watershed. The test results with model efficiencies (E = 0.65 - 0.70 for daily outflow, 0.49 - 0.94 for daily water table, and 0.57 - 0.61 for soil respiration) showed that the linkage is applicable for modeling C and N dynamics in the watershed. The simulation results also showed that this watershed is a significant C sink, with an average C sequestration rate of about 2.2 Mg C/ha/yr for the period of 2003-2007. However, the C sequestration in this watershed was substantially affected by the climate conditions. For example, the predicted annual C storages for 2004 and 2007 were over 45% lower than those for the other years because of drought; the precipitation in 2004 and 2007 was about 390 and 430 mm less, respectively, than the long-term average (1350mm). The results also showed that there was a significant relationship between annual soil heterotrophic CO2 emission and annual average water table depth (R2 ranging from 0.93 to 0.99 for each cell, P<0.01). High water table levels led to low soil CO2 but high methane emissions from the watershed. The predicted annual methane emission rates proportionally increased with precipitation. However, severe drought may cause a reduction in methane emission by over 80%, but an increase in CO2 emission by more than 50%. In addition, the average N2O emission from the watershed was 3.7 kg N/ha/yr in 2003-2007, and the annual emission was significantly decreased with increasing precipitation (R2 = 0.93, P<0.01). The predicted NEE fluxes showed substantial variations in space and time. The spatial variation was mainly due to the differences in vegetation and water table associated with topography across the watershed, and the temporal variation was primarily caused by climate. The spatial pattern of methane fluxes follows the topography.