Application of TRIPLEX model for predicting Cunninghamia lanceolata and Pinus massoniana forest stand production in Hunan Province, southern China

The process-based hybrid model is a promising tool for predicting forest stand production on regional scales. TRIPLEX1.6 was adapted and parameterized to simulate Cunninghamia lanceolata and Pinus massoniana forest stand production in Hunan Province, China, using data from permanent sample plots established by the National Forest Inventory of China (CNFI). Monthly maximum and minimum air temperature and precipitation (derived from interpolation of the data collected at 369 meteorological stations in Hunan between 2000 and 2009) were used to run the model. Model calibrations and simulations were performed through threshold parameters and initial statuses at a regional scale. The species- and site-specific sensitive parameters were adjusted for estimating tree growth rate of different stand age or diameter at breast height (DBH). The improved parameterize procedure actually did increase model practicability. The site and species data for model validation were achieved by applying half the 2009 permanent sample plot data. Estimated stand average tree height (H), DBH, and biomass were validated against the other half of the 2009 data. Simulated results were consistent with the observed data in Hunan Province. Coefficients of determination (r2) of predicted and observed data were 0.83 for H, 0.82 for DBH, 0.90 for aboveground biomass, and 0.94 for total biomass, indicating that TRIPLEX1.6 is capable in predicting forest growth and biomass dynamics of subtropical coniferous forests. Moreover, independent validations determined that TRIPLEX1.6 demonstrated competence in extrapolating outcomes on regional scales as well as withstanding rigorous testing in predicting C storage in subtropical forest ecosystems.

► TRIPLEX was used to simulate production of subtropical coniferous forests in China.
► We calibrate and validate model with threshold and adjusted parameters.
► We simulate and analyze the spatial pattern of LAI, NPP and biomass C.
► Improving parameterization will increase the model practicability and performance.
► Stand density and age show superior spatial impacts on forest stand growth.