ETKF-3DVAR mixed-assimilation Project types: National natural science foundation of China (general program) Team leader: Yin Ren Period of Project: 2015 .1-2018 .12 Spatial and temporal scaling up of carbon fixation in Eucalyptus plantations enables precise estimation of the carbon sequestration capability and its potential effect on the quantitative assessment on a regional scale. The purpose of this study is thus to demonstrate a means of estimating carbon fixation accurately, and to decrease the uncertainty of simulation results using mixed-assimilation and 3-PG2 model direct extrapolation. This is done by applying the real value calculated at a single plot scale to a regional scale. The project team have previously established that the largest canopy conductance (gCx) along with another 8 parameters, are the dominant factors that influenced the outcomes of 3-PG2 models for simulating biomass at plot scale. However, optimal parameter estimation is still unclear during the process of up scaling, and it lacks a quantitative analysis methodology for nonlinear systems. In this study, which takes Zhangzhou (a region of Fujian Province where Eucalyptus plantations are widespread) as a focus area) we confirm key parameters in a 3-PG2 model using a Fourier Amplitude Sensitivity Test and Bayesian method. We introduce a Weather Research and Forecast (WRF) model to improve simulation accuracy based on the combination of multi-scale observation data collection (as is done in the Ensemble Transform Kalman Filter (ETKF)) in combination with use of the Three Dimensional Variational (3DVAR) data assimilation algorithm.Scaling up of carbon fixation is used by direct extrapolation through 3-PG2 model optimization parameters. Finally, we predict the spatial and temporal dynamics of carbon fixation in Eucalyptus plantations under different management systems and climate change scenarios.