Forest carbon storage in China from 2003 to 2021: Estimation based on the volume-derived carbon storage model with scale-compatible and tree species-merged

Abstract

Accurately estimating and assessing the forest carbon storage in China is crucial to achieving the carbon peak and neutrality targets. In this paper, utilizing data from 173,031 plots obtained during China's 6–9th national forest inventory, we established a volume-derived forest carbon model based on scale-compatible and tree species-merged method. Utilizing the statistical data of forest area and volume from 2003 to 2021, we estimated and evaluated the national forest carbon storage, and analyzed the spatial and temporal distribution of forest carbon. We also analyzed the uncertainty of national forest carbon storage and recommended a more suitable forest carbon storage estimation model. Our study indicated that: (1) Determination coefficients (R²), standard error of estimate (SEE) and mean prediction error (MPE) for the population average model were 0.964, 5.85 t/ha, and 0.60 %, respectively. The model we have established was better and can be applied to estimate and evaluate the national forest carbon storage. (2) When all plots were used as modeling units, the R², SEE and MPE of the scale-compatible model were 0.914, 10.69 t/ha and 0.13 %, respectively, while the tree species-merged method had a better fit, with the R², SEE and MPE of 0.960, 7.26 Mg ha⁻¹ and 0.09 %, respectively. (3) From 2003–2021, the forest carbon storage in China increased rapidly, from 4.96 ± 0.25 Pg C in 2003–7.95 ± 0.40 Pg C in 2021, with an average annual carbon sink of 0.166 Pg C/yr. Forest carbon storage was mainly distributed in the southwest and northeast, and was also mainly stored in Pinaceae. It is worth noting that from 2003 to 2013, hard broadleaved forest played a role as a carbon source due to the sharp decrease in area. (4) The average uncertainty of national forest carbon storage estimated by the tree species-merged method was 5.05 %, and the minimum was only 5.03 %. Compared with the most detailed model (a provincial-scale model with distinguishing dominant tree species group and stand characteristics), the estimation error of forest carbon storage in the simplest model (a national-scale model without distinguishing forest level and stand characteristics) was less than 5 %. Therefore, we recommended that the simplest model can be used to estimate the national forest carbon storage. The forest carbon storage model proposed in this paper can provide support for accurately estimating and evaluating forest carbon storage in China.