Carbon Fluxes Over An Estuarine Wetland: In Situ Measurement And Modeling

With the help of eddy covariance and remote sensing technique, we investigated the carbon dynamic patterns and the associated driving factors and also tried to simulate gross primary productivity of an estuarine wetland located on Chongming Island of Shanghai, China. Meanwhile, we examined how plant invasion and tidal activity affect wetland carbon fluxes. The main findings are summarized as follows:(1) During three years observation (2005-2007), the wetland behaved as strong C-CO2 sink, with the strength of 562-834 g C m-2 yr-1. Seasonal patterns of net ecosystem exchange of CO2 (NEE) showed significant differences between high and low elevation sites in 2005 and 2007, and the cumulative NEE at high elevation site was higher than that at low elevation site by a magnitude of 100 g C m-2 yr-1. The carbon fluxes showed similar responses to meteorological variables at two sites, but differed largely in extreme conditions. The decreased tidal activity along the altitudinal gradient, combined with the different plant speceis composition, mainly accounted for the different behaviors of carbon fluxes. This study identified the heterogenity of carbon fluxes and their response to meteorological variables inside ecosystems with environmental gradients, which brought great uncertainty in whole ecosystem estimation and upscaling.(2) One year of continuous data from high and low elevation sites were analyzed to evaluate the tidal effect on carbon flux. The measured wavelet spectra and cospectra of NEE and other environmental factors demonstrated that the dynamics of NEE at both sites exhibit a tidal-driven pattern with obvious characteristics at scales between 10 and 20 days (256-512-h). Environmental factors exerted major controls on the carbon balance at finer temporal scales. NEE was more sensitive to tides at the low elevation site than at the high elevation site. Overal the mean nighttime NEE during spring tides was lower than that during neap tides, indicating suppressed ecosystem respiration under inundation. Larger differences were observed at the low elevation site due to the longer inundation. In contrast, daytime NEE was more variable since plant reacted differently in different growth period and under different tidal elevations. Whilst tides would also transport organic matter to nearby estuary and hence may incur carbon emission in the receiving ecosystems. Thus, further study on lateral carbon transport is required to investigate the tidal effect on the carbon sink/source role of the wetland.(3) From 2005 to 2006, the percent of cordgrass (Spartina alterniflora) increased by 10-30% on the wetland, and largest increases were observed when the previous percent fell in the range of 20-40%. For each site, the fetch that faced the sea experienced larger increases than that faced the dike. Gross primary production (GPP), ecosystem respiration (RE) and net ecosystem productivity increased with rising percent of S. alterniflora. The increased carbon sequestration rates in C3 species community invaded by C4 species S. alterniflora showed potential in alleviating climate warming. The invasion of S. alterniflora also alterd the responses of carbon flux to environmental factors. With greater percent of S. alterniflora, the light inhibition under strong insolation was relieved; the temperature sensitivity increased when temperature was below 26℃, while increased when temperature rose above 26℃.(4) Vegetation photosynthesis model (VPM), which combines the measurements of eddy flux tower and satellited-derived data, shows robust in simulating ecosystem productivity and thus acts as a strong candidate for upscaling work. We tried two methods to calculate maximum light use efficiency (e0), which in large part determined the performance of VPM. e0 derived from long-term (almost whole growing season) dataset would lead to an underestimate of GPP. In comparison, use of short-term eo greatly enhanced the simulation. This was mainly due to the summer passive condition and scattered distribution of data points for estimating eo. Meanwhile, uptake of CO2 by canopy was more like instantaneous response. Best simulation was observed at high elevation site. We proposed that the tidal activity affected the reflectance and associated vegetation indices (VIs), which was partly reflected in the dynamics of VI along the altitudinal gradient. Thus, a refinement of VI is required for ecosystems that experience flooding. Some other reasons, like bias in the partitioning of NEE into GPP and RE, different weather conditions on days that satellite does not pass by was also discussed.