Background The distribution pattern and estimation model of forest biomass differ greatly in different regions. It is necessary to establish a biomass growth model suitable for specific area and specific environmental conditions to accurately estimate regional forest biomass and reduce estimation errors. To establish biomass allometry models of different forest types in Shanxi Plateau is of scientific significance for forest productivity and carbon cycle research.

Methods Based on the data of a total of 97 trees of 9 tree species, which was grouped into deciduous broad-leaved forest(DBF), temperate coniferous forest(TCF) and cold temperate coniferous forest(CCF), from 10 representative ecological stations in Shanxi province, as well as their biomass data of the each part of the species measured by harvest method, we established their biomass regression models by using diameter at breast height(D) and height(H) of a tree as the independent variables for the three forest types to compare the accuracy of the models.

Results 1) The overall average proportion of total biomass in trunk, crown and root of the 9 tree species was 47.22%, 25.59% and 27.18%, respectively; except for Pinus sylvestris, the tree trunk biomass accounted for the largest proportion of the total biomass. The ratio of root to shoot of the all tree species ranged from 0.12-0.88, with a maximum and minimum ratio in both Platycladus orientalis and Pinus sylvestris. 2) Among trunk biomass models, the model including both D and H variables fitted well for all the species. The determination coefficient(R²) of the model for DBF, TCF and, CCF species, when both D and H was used as a combining variable(D²×H), reached 0.82, 0.97 and 0.96, respectively. The R² of the model, when D and H were used as additive variable(D+H), were 0.84, 0.97 and 0.98, respectively, for DBF, TCF and CCF, respectively. For canopy and root biomass, the performance of the models was similar to those for the trunks, and both D and H could be well used to estimate their biomass. The R² values of the fitted models for canopy biomass were 0.67, 0.75 and 0.88 for DBF, TCF, and CCF, respectively, and 0.61 and 0.70 for the roots biomass for both DBF and CF. 3) Comparing with the results from the Chinese Forest Model(CFM), the biomass prediction deviation from the current models was significantly less than those from the CFM.

Conclusions The established allometric equations using both D and H as independent variables in our study are more suitable for biomass estimation of this area. In practical applications, a model that is applicable and meets the accuracy requirements should be selected for different sites.