Optimization of water uptake and photosynthetic parameters in an ecosystem model using tower flux data

The soil water stress factor (f(w)) and the maximum photosynthetic carboxylation rate at 25 degrees C (V-cmax) are two of the most important parameters for estimating evapotranspiration and carbon uptake of vegetation. Ecologically these two parameters have different temporal variations and thus their optimization in ecosystem models poses a challenge. To minimize the temporal scale effect, we propose a three-stage approach to optimize these two parameters using an ensemble Kalman filter (EnKF), based on observations of latent heat (LE) and gross primary productivity (GPP) fluxes at three flux tower sites in 2009. First, the EnKF is applied daily to obtain precursor estimates of V-cmax and f(w). Then, V-cmax is optimized at different time scales, assuming f(w) is unchanged from the first step. The best temporal period is then determined by analyzing the coefficient of determination (R-2) of GPP and LE between simulation and observation. Finally, the daily f(w) value is optimized for rain-free days corresponding to the V-cmax curve from the best temporal period. We found that the variations of optimized f(w) are largely explained by soil water content in the summer. In the spring, the optimized f(w) shows a smooth increase following the rise of soil temperature, indicating that f(w) may respond to the development of fine roots, which is related to the amount of accumulated heat in the soil. The optimized V-cmax generally follows a pattern of a rapid increase at the leaf expansion stage in the spring, small variation in summer, and an abrupt decrease at foliage senescence. With eddy covariance fluxes data, data assimilation with a EnKF can retrieve the seasonal variations of water uptake and photosynthetic parameters in an ecosystem model, and such gives clues on how to model forest responses to water stress. (C) 2014 Elsevier B.V. All rights reserved.