Objective This paper aims to establish a compatible biomass model system for individual tree components of Cunninghamia lanceolata, Pinus massoniana, and Quercus species, so as to provide technical support for the scientific and standardized modeling of main tree species in Hunan Province of central China and the accurate prediction of forest biomass.

Method Based on the felling data of 468 individual trees of three main tree species in Hunan Province (160 of Cunninghamia lanceolata, 153 of Pinus massoniana, and 155 of Quercus species), the basic models of biomass for six individual tree components (aboveground, underground, trunk, bark, branch, and leaf biomass) were compared respectively, and the optimal model with the best fitting indicators was sought. Further, the biomass models of different origins (natural forests and plantations) were compared. Dummy variable models were established by selecting parameters adding origin as variable with significant differences, and the biomass of individual tree components was predicted by constructing nonlinear simultaneous equations with proportion control.

Result The fitting results of basic biomass models for individual tree components of three tree species showed that for four components of aboveground biomass B₁, underground biomass B₂, trunk biomass B₃, and bark biomass B₄, the independent variables selected were DBH and tree height, while for the biomass models of branches B₅ and leaves B₆, DBH, tree height (H), and crown width (CW) were selected. The models with origin as a dummy variable indicated that for aboveground biomass B₁ (the mean prediction error MPE was reduced by 1.16% compared with the base model) of Cunninghamia lanceolata, underground biomass B₂ (MPE was reduced by 4.65%) of Pinus massoniana, and aboveground biomass B₁ (MPE was reduced by 3.24%) and underground biomass B₂ (MPE was reduced by 29.13%) of Quercus species, the biomass in natural forests was significantly higher than that in plantations. The fitting effect of individual tree biomass model system with origin as a dummy variable was better, with the determination coefficients (R²) of aboveground biomass models of three tree species all above 0.90, and the average predicted error MPE was below 4%, while the underground biomass models also showed R² above 0.90 and MPE below 14%.

Conclusion The individual tree component biomass model with origin as a dummy variable has the best fitting effect, followed by the nonlinear simultaneous equation compatibility model system, and the basic model has the worst fitting effect. Although the fitting effects of model systems are slightly worse than the dummy variable model, these model systems can solve the compatibility and additivity of individual tree component biomass, and the differences among their parameters are not significant. Therefore, it is recommended to use compatibility model systems to predict the biomass of individual tree components of Cunninghamia lanceolata, Pinus massoniana, and Quercus species in Hunan Province of central China.