Date of Award
Doctor of Philosophy (PhD)
Earth, Atmospheric, and Planetary Sciences
Committee Member 1
Committee Member 2
Devdutta S. Niyogi
Committee Member 3
Terrestrial ecosystem plays a critical role in the global carbon cycle and climate system. Therefore, it is important to accurately quantify the carbon dynamics of terrestrial ecosystem under future climatic change condition. This dissertation evaluates the regional carbon dynamics by using upscaling approach, mechanistically-based biogeochemistry models and in situ and remotely sensed data.
The upscaling studies based on FLUXNET network has provided us the spatial and temporal pattern of the carbon fluxes but it fails to consider the atmospheric CO2 effect given its important physiological role in carbon assimilation. In the second chapter, we consider the effect of atmospheric CO2 using an artificial neural network (ANN) approach to upscale the AmeriFlux tower of net ecosystem exchange (NEE) and the derived gross primary productivity (GPP) to the conterminous United States. We found that atmospheric CO 2effect on GPP/NEE exhibited a great spatial and seasonal variability. Further analysis suggested that air temperature played an important role in determining the atmospheric CO2 effects on carbon fluxes. In addition, the simulation that did not consider atmospheric CO2 failed to detect ecosystem responses to droughts in part of the US in 2006. The study suggested that the spatially and temporally varied atmospheric CO2 concentrations should be factored into carbon quantification when scaling eddy flux data to a region.
The process-based ecosystem models are useful tools to predicting future change in the terrestrial ecosystem. However, they suffer the great uncertainty induced by model structure and parameters. The carbon isotope (13C) discrimination by terrestrial plants, involves the biophysical and biogeochemistry processes and exhibits seasonal and spatial variations, which may provide additional constraints on model parameters. In the third chapter, we found that using foliar 13C composition data, model parameters were constrained to a relatively narrow space and the site-level model simulations were slightly better than that without the foliar 13C constraint. The model extrapolations with three stomatal schemes all showed that the estimation uncertainties of regional carbon fluxes were reduced by about 40%.
In addition, tree ring data have great potentials in addressing the forest response to climatic changes compared with mechanistic model simulations, eddy flux measurement and manipulative experiments. In the fourth chapter, we collected the tree ring isotopic carbon data at 12 boreal forest sites to develop a linear regression model, and the model was extrapolated to the whole boreal region to obtain the water use efficiency (WUE) and GPP spatial and temporal variation from 1948 to 2010. Our results demonstrated that most of boreal regions except parts of Alaska showed a significant increasing WUE trend during the study period and the increasing magnitude was much higher than estimations from other land surface models. Our predicted GPP by the WUE definition algorithm was comparable with site observation, while for the revised light use efficiency algorithm, GPP estimation was higher than site observation as well as land surface model estimates. In addition, the increasing GPP trends estimated by two algorithms were similar with land surface model simulations.
Liu, Shaoqing, "Quantifying terrestrial ecosystem carbon dynamics with mechanistically-based biogeochemistry models and in situ and remotely sensed data" (2016). Open Access Dissertations. 968.