Using Nolinear Mixed Model And Dummy Variable Model Approaches To Analyzing Parameters Of Height-age Models For Pinus Armandi Franch Provenances

Height of Pinus armandi Franch in provenance test forest wasmeasured at ages4to10,24and32in Dalaoling Forest Farm,Yichang County,HubeiProvince. This study used nonlinear mixed model and dummy variable modelmethods to Analyzing Parameters of Height-Age Models for Pinus armandi FranchProvenances. The conclusion of two subject-specific approaches was that at a givenlocation, the asymptote parameter of the height-age curve differs by seed sources. Butthe shape parameter and rate parameter of the height-age curve was the same amongseed sources. Models and test precisions had been analyzed and compared with5indicators.they were determination coefficient(R~2),residual sum of squares(SSE),Minimum root mean square error(RMSE) and absolute residualsmean(|E|).Linear mixed model with random parameters and dummy model withspecific(1oca1) parameters were better than population average model. average SSE,average|E|, average RMSE and average R~2of two subject-specific approaches were18.27%,12.77%,9.6%lower and1.37%higher than population average modelï¼›SSE,|E|, RMSE and R~2of dummy variable model were1.8%,4%,0.9%lower and0.1%higher than nonlinear mixed model. By test precisions of three models,wechoosed dummy variable model as Generalized Height-Age Models for Pinusarmandi Franch Provenances, SSE,|E|, RMSE and R~2of dummy variable model were19%,14.57%,10%lower and1.42%higher than population average model..