Estimation And Uncertainty Analysis Of Grassland Ecosystem Productivity In Northern China

As the widest distributed ecosystem type, grassland play an important role in global carbon cycle. Accurate estimation of grassland productivity is crucial for the sustainable use of grassland resources. In this study, the spatial grassland biomass and gross primary productivity (GPP) across Northern China during 2001-2013 were assessed based on field measurements, the eddy covariance flux data, remote sensing data and the meteorology interpolation data. Moreover, the relative uncertainty of regional grassland biomass and GPP was analyzed. And the potential sources of uncertainty, including remote sensing data sources, model forms and model parameters, were determined and their relative contribution was quantified. Main findings are:(1) The temporal and spatial distribution characteristics of grassland biomass in Northern China:The results showed that the annual grassland biomass in Northern China was 548.5±130.5 g m-2 (i.e.,1248.6±297.2 Tg) during 2001-2013, increasing in recent years. And the grassland biomass showed an increasing pattern from western to eastern area. Different grassland types held different biomass density. Alpine steppe and alpine meadow in the northwest possessed lower productivity than temperate meadow and temperate steppe in the medium part. And montane meadow in the east held the highest productivity. Generally, the higher temperature, higher precipitation could result in the higher grassland biomass density in the southeast.(2) The uncertainty analysis of grassland biomass in Northern China:The mean relative uncertainty in Northern China grassland biomass was 19.8%, the higher value occurring in the lower grassland biomass area. The alpine steppe held the larger relative uncertainty (20%-40%), while the temperate or tropical steppe held the minimum value of relative uncertainty in grassland biomass estimation. There were distinguishable differences among the uncertainty contributions of three sources (model parameters> model input> model forms), which contributed 52%,27% and 13%, respectively. This study highlighted the need to concern the uncertainty in grassland biomass estimation, especially for the uncertainty related to model parameters.(3) The temporal and spatial distribution characteristics of grassland GPP in Northern China:The results showed that the annual GPP in Northern China was 833.3 Tg (i.e.,349.5 g C m-2) during 2001-2013, increasing in recent years. And the grassland GPP showed an increasing pattern from western to eastern area. Different grassland types held different productivity. Alpine steppe in the northwest possessed lower GPP, while tropical meadow and montane meadow in the medium and east part held higher productivity.(4) The uncertainty analysis of grassland GPP in Northern China:The mean relative uncertainty in Northern China grassland GPP was 38%, while most pixels held value of<30%, little with 50%-70%, which suggested that the GPP results were significantly different among the different GPP models. Similarly, the higher value occurring in the lower GPP area, explained by the model data themselves. If the model is very sensitive to an input with very small uncertainty, then the uncertainty contribution of this input might be small, and vice versa. The model forms had a crucial impact on the estimation of regional GPP.This study focused on the quantitative analysis of regional uncertainty, providing a creative factorial style for the accurate estimation of grassland productivity in Northern China.