Characteristic Of CO₂Flux In Wetland Ecosystem And Gap-filling Strategy

Observation of carbon dioxide (CO2) flux between wetland ecosystem and atmosphere is an important way to estimate the amplitude of carbon source/sink, and also provides foundation knowledge to reveal the global carbon cycle. The summaries of CO2flux dynamics characteristics and the relationships established between CO2flux and environmental factors are all base on those observation data. These information are curial in understanding and predicting what wetland is going to be in future global change.Eddy covariance technology is the most widely used flux observation technology. The advantage of this technology is that it provides a continuous direct measurement at ecosystem scale. Meanwhile, the shortcut is lots of data gaps that failed to pass quality test. Before these observation data were involved in dynamic analysis, some specific gap-filling algorithms would be applied to fill those gaps. One of deficiencies of this approach is that (1) the current gap-filling algorithms are mostly based on mathematics rather than ecological understandings. The other deficiency is that (2) the results of dynamic analysis of gap-filled data are not reliable, because those data have already been contaminated by those gap-filling algorithms. Although some important characteristics of CO2flux have already been obtained by time series analysis methods, these information were rarely used to improve gap-filling algorithms.Vegetation phenology is also responsible for seasonal changes of CO2flux. Currently, auxiliary observations of eddy covariance technology didn’t include vegetation phenology observation. Introducing the satellite-based vegetation index would be a solution.This study used the observation data from two flux sites in Chongming Dongtan to investigate the different characteristic of CO2flux in diurnal and annual cycle and the relationship between those characteristic and environmental factors and vegetation phenology. Subsequently, these information were applied to improve the gap-filling algorithm. The results showed that (a) in diurnal cycle, the photosynthetically active radiation (PAR), air temperature (Ta) and vapor pressure deficit (VPD) all showed highly correlations with CO2flux in daytime. However, time-lag analysis showed that PAR was the only factor leading or synchronize with CO2flux, while Ta and VPD showed a delay. We can infer that PAR was the major factor which influenced the fluctuations of daytime CO2flux.In comparison, nighttime CO2flux didn’t show any significant correlation with these factors.(b) In annual cycle, average nighttime temperature showed a high correlation with CO2flux, which meant temperature were highly correlated with ecosystem respiration at relatively longer time scales. Results of light response equation regression of daytime CO2flux showed that vegetation indices were highly correlated with ecosystem light use efficiency (alpha), while average daytime temperatures were highly correlated with ecosystem optimum gross production rate (GPPopt).(3)The plant invasion happened in Dongtan wetland prompted temperature sensitivity of ecosystem respiration and light use efficiency of ecosystem photosynthesis, and ecosystem responses to environmental factor change were also be improved.The high coefficients of determination between gap-filled data and observation values showed that the gap-filled data simulated the observation values successfully. But the larger relative root mean square errors (rRMSE) and low slopes illustrated that this gap-filling algorithms still need to be improved by:(1) using multi-step correlate model to simulate ecosystem respiration dynamic in diurnal cycle;(2) integrating vegetation variations to better simulate ecosystem respiration in annual cycle;(3) validating seasonal change of EVI;(4) introducing accessory factors into gap-filling algorithm.This study found that some characteristics of CO2flux dynamic may also exist in other ecosystems:(1) changes in canopy conductance may explain the delay between PAR and CO2flux;(2) light use efficiency and GPPopt may show delay in mixed communities in annual cycle.