Sensitivity and Uncertainty Analyses of Flux-based Ecosystem Model towards Improvement of Forest GPP Simulation

Sensitivity Analysis of Biome-BGC for Gross Primary Production of a Rubber Plantation Ecosystem: A Case Study of Hainan Island, China

Accurate monitoring of forest carbon flux and its long-term response to meteorological factors is important. To accomplish this task, the model parameters need to be optimized with respect to in situ observations. In the present study, the extended Fourier amplitude sensitivity test (eFAST) method was used to optimize the sensitive ecophysiological parameters of the Biome BioGeochemical Cycles model. The model simulation was integrated from 2010 to 2020. The results showed that using the eFAST method quantitatively improved the model output. For instance, the R² increased from 0.53 to 0.72. Moreover, the root-mean-square error was reduced from 1.62 to 1.14 gC·m^-2·d^-1. In addition, it was reported that the carbon flux outputs of the model were highly sensitive to various parameters, such as the canopy average specific leaf area and canopy light extinction coefficient. Moreover, long-term meteorological factor analysis showed that rainfall dominated the trend of gross primary production (GPP) of the study area, while extreme temperatures restricted the GPP. In conclusion, the eFAST method can be used in future studies. Furthermore, eFAST could be applied to other biomes in response to different climatic conditions.

Uncertainty analysis of ecosystem services and implications for environmental management - An experiment in the Heihe River Basin, China

Integrating the economic values of ecosystem services provided by different land uses into decision analysis is critical to achieving effective environmental management in endorheic basins. However, policymaking often ignores the uncertainty related to the variability of parameters in ecosystem service values. To this end, we identified sensitive parameters in the ecosystem service values under four land uses using the global sensitivity analysis method and quantified the potential monetary outcomes based on the Monte Carlo method. The results indicated that only a few sensitive parameters, such as water yield (Q_i) and treatment costs per unit of nitrogen (Cost_N), were the primary sources of uncertainty. Therefore, we suggest that improving the precision of sensitive parameters is essential for reducing uncertainty in the total ecosystem service value. Additionally, the overall monetary outcomes for cropland exhibited negative values and had higher risk and lower benefits than those for forest from the standpoint of ecosystem services. In addition, the nonmarketed service of landscape aesthetic made the monetary outcomes of water bodies higher than those of cropland, yet the value of landscape aesthetic was highly uncertain. Therefore, efforts should be made to improve total monetary outcomes by decreasing the negative values in food provisioning of cropland and the uncertainty in landscape aesthetic for water bodies. The sensitivity analysis and uncertainty analysis provide important guidelines for quantifying and reducing the related uncertainty and provide policy information for environmental management based on a comprehensive consideration of the potential ecosystem service values for various land uses.