A Three-Level Model System of Biomass and Carbon Storage for All Forest Types in China

• Published in: Forests 2024-08, Vol.15 (8), p.1305
• Main Authors: Zeng, Weisheng, Zou, Wentao, Chen, Xinyun, Yang, Xueyun
• Format: Article
• Language: English
• Subjects: Accuracy; Analysis; Biomass; biomass conversion factor; Biomass energy; Carbon; Carbon sequestration; carbon storage; Climate change; Comparative analysis; dummy variable; error-in-variable; Errors; Estimates; Forest biomass; Forest carbon; Forest ecology; Forest resources; Forests; National forests; Plant species; Regional development; Regression analysis; Regression models; Simultaneous equations; Storage; Trees; Variables
• ISSN: 1999-4907; 1999-4907
• DOI: 10.3390/f15081305

Forest biomass and carbon storage models are crucial for inventorying, monitoring, and assessing forest resources. This study develops models specific to China’s diverse forests, offering a methodological foundation for national carbon storage estimation and a quantitative basis for national, regional, and global carbon sequestration projections. Utilizing data from 52,700 permanent plots obtained during China’s 9th national forest inventory, we calculated biomass and carbon storage per hectare for 35 tree species groups using respective individual tree biomass models and carbon factors. We then constructed a three-level volume-based model system for forest biomass and carbon storage, applying weighted regression, dummy variable modeling, and simultaneous equations with error-in-variables. This system encompasses one population of forests, three forest categories (level I), 20 forest types (level II), and 74 forest sub-types (level III). Finally, the assessment of these models was carried out with six evaluation indices, and comparative analyses with previously established biomass models of three major forest types were conducted. Determination coefficients (R2) for the population average model, and three dummy models on levels I, II, and III, exceed 0.78, 0.85, 0.92, and 0.95, respectively, with corresponding mean prediction errors (MPEs) of 0.42%, 0.34%, 0.24%, and 0.19%, and mean percent standard errors (MPSEs) of approximately 22%, 21%, 15%, and 12%. Models for 20 forest types and 74 sub-types yield R2 values above 0.87 and 0.85, with MPE values below 3% and 5%, respectively. Notably, the estimates of previous biomass models of three major forest types demonstrated considerable uncertainty, with TRE ranging from −20% to 74%. However, accuracy has improved with larger sample sizes. In total biomass and carbon storage estimations, the R2 values of dummy models for levels I, II, and III progressively increase and MPSE and MPE values decrease, whereas TRE approximates zero. The tiered model system of simultaneous equations developed herein offers a quantitative framework for precise evaluations of biomass and carbon storage on different scales. For enhanced accuracy in such estimations, applying level III models is recommended whenever feasible, especially for national estimation.