| The researches related to forest biomass and biomass growth are key areas and key point in the research of today.In this study,the pinus densata natural forest was treated as the research object in Shangri-La city of Yunnan province,the individual tree wood biomass growth models and stand biomass components growth models were constructed based on the individual tree wood biomass data and stand wood biomass data.The plot effect and tree effect were considered as random effect,the individual tree wood biomass growth models were constructed by nonlinear mixed effect model technology,and all the different random parameter combinations were fitted and the variance and covariance structures of the models were analyzed.Secondly,based on the five theoretical growth equations,I constructed the stand biomass components growth models.Then,based on the dynamic variation of the stand diameter structure,the diameter cumulative structure model and diameter distribution dynamic forecasting model were built.Combined the former two models with individual tree biomass components models and height-diameter model,the stand biomass components were predicted and compared the predictions with the optimal theoretical growth models.The results showed that:(1)Based on the theory growth equation,the wood biomass growth model was built.The Richards model had the best performance,the determination coefficient R~2was 0.889,the root mean square error RMSE was 92.088;The mixed effect models improved the fitting precision comparing with the Richards model.The two-level mixed effect models had the highest prediction accuracy and the values reached 93.05%,the Richards model was only 93.00%.(2)Based on the five theories of the growth equation,the stand biomass components growth models were constructed.The determination coefficient R~2 of optimal models for the growth of the stand biomass of wood,bark,leaves,branches,and aboveground were 0.655,0.583,0.401,0.561,0.611 respectively.The root mean square error RMSE were 24.198,4.066,1.383,4.072 and 34.849.The estimated accuracy P were72.962%,76.921%,65.551%,82.623%and 77.006%.(3)Based on the variable Richards model,the cumulative structure model of stand diameter was constructed.The model obtained a good fitting effect.The determination coefficient(R~2)of the equations were all over 0.9 and the mean value of the root mean square error(RMSE)was 0.427 in 84 plots.(4)Made the parameters dynamic estimate models of the stand cumulative diameter distribution equations by 1stOpt and both the parameter prediction models obtained better model fitting and prediction performance.The determination coefficient(R~2)of the b parameter and the c parameter reached 0.883 and 0.874 and the root mean square error(RMSE)were 0.105 and 25.203 respectively.The estimated accuracy P were77.625%and 69.558%respectively.(5)Based on power function model,the individual tree wood biomass models for the wood,bark,leaves,branches and aboveground were constructed.The optimal model of determination coefficient R~2 were 0.990,0.898,0.674,0.831,0.990.The root mean square error RMSE were 29.320,11.175,4.389,18.989 and 36.145 respectively.The estimated accuracy P were 88.815%,87.686%,72.422%,76.156%,89.069%respectively.(6)Based on five basic models,the height-diameter models were constructed.The optimal model Logistic determination coefficient R~2 was 0.877,the root mean square error RMSE was 2.349 and the estimated accuracy P was 92.772%.(7)Compared with general stand biomass growth model,the wood,leaves and aboveground models of stand biomass growth model showed the higher forecast accuracy while the forecast accuracy of general stand biomass growth models in bark and branches biomass were better. |
PDF Full Text Request