Comparing estimation algorithms for compatible biomass models of Moso Bamboo.

Moso bamboo (Phyllostachys edulis) forest is one of the important alternative soruces in the tropical and subtropical regions. Its biomass quantification can be a basis of carbon storage estiamtion of the bamboo forest. Based on the destructively sample measurements of the moso bamboo in the Jiangsu Province in China, three forms of the model structures (controlling by sum of equation (CSE), level-1 proportional fitting distribution (L1PFD) and level-2 proportional fitting distribution (L2PFD)), and three estimation algorithms (ordinary least squares (OLS), two stage least square (2SLS) and three stage least square (2SLS)) were evaluated for identifying the best compatible biomass model. Results showed that all the three model structures solved the compatibility problems effectively. With each estimation algorithm applied, difference of the fit statistics between L1PFD and L2PFD was insignifcant. The fit statistics produced by 2SLS were slightly different from those of OLS, but fitting accuracy the latter was slightly higher, and had a more efficient estimation procedure. Thus, we recommend L1PFD model for predicting biomass for moso bamboo with use of diameter at breast height as a predictor variable. Our model will provide a reliable basis for estimating biomass and carbon stock of the moso bamboo forest in China. [ABSTRACT FROM AUTHOR]