关键词: 异速生长方程, 理论生长方程, 木荷, 生物量, 广东省, 动态预测

Abstract: [Objective] In order to predict the biomass dynamic for Schima superba in Guangdong, a method of model system of combining the theoretical growth equation and the allometric equation was proposed and fitted. The method will provide the method ological support to other tree species for measuring carbon sink.[Method] Based on the measuring biomass survey data of sample trees, including 40 large trees whose stem were analyzed, this paper established the model system from theoretical growth equation which related DBH (diameter at breast height) to age and allometric equation which related aboveground biomass to DBH. Using nonlinear error-in-variable simultaneous equations, the models were fitted and the parameters were estimated under the classificated parameters which described DHB' growth rate. Based on three periods data of the sample trees in the permanent sample plot in China national forestry inventory, biomass dynamics for Schima superba were predicted in Guangdong province. Determination coefficient (R²) and root mean square error (RMSE) were used to evaluate the model and estimate error for biomass storage and increment were used to assess the prediction.[Result] By classificated parameters for DHB growth, the explanation of theoretical growth equation for DBH was more than 0.95, raised 0.166 3 than that of the model with fixed growth parameter, RMSE reduced above 2.16 to 1.97 cm. The explanation of aboveground biomass with predicted DBH was 0.82, improved 0.30 than that of the model with fixed growth parameter, closely to R² of independent allometric model, and RMSE reduced more than 30 to 51 kg. The estimated errors for different biomass storages ranged from -46.31% to 77.45% by the model with fixed growth parameter, while the estimated errors reduced from -16.13% to -7.06% in the case of the model with classificated parameters. There was same law at different scales for the two models, the estimated error for single tree were lower than that for stand and the estimated error for stand was lower than that for region. The error difference between single tree and region were not greater than 10% by the model with fixed growth parameter, while it was lower than 8% by the model with classificated parameters. The estimated errors for biomass increment at different periods were generally larger which varied from 32.57% to 115.45% by the model with fixed growth parameter, while the errors reduced from -6.57% to 15.77%,even less than ±10% at tree level, by the model with classificated parameters. With enlarging of scale, the estimated errors increased, the error differences between single tree and region varied from 10% to 15% by the model with fixed growth parameter and the differences were relatively stable around 8% by the model with classificated parameters.[Conclusion] By means of the combination model system of theoretical growth equation and allometric equation, classification could significantly improve the model accuracy and reduce the estimated error for prediction. Only DBH or age in the case of the model with fixed growth parameter, two-stages DBH or DBH and age of same period in the case of with classificated parameters could be used to predict the biomass dynamic in the future. The method and models were easy to use and had a promising applicative value to the estimation of carbon sink national forestry inventory and afforestation of carbon sink,the estimated error at regional scale could basically satisfy the accuracy requirements.

Key words: allometric growth equation, theoretical growth equation, Schima superba, biomass, Guangdong Province, dynamic prediction