Uncertainty Assessment For Regional-level Forest Above-ground Biomass Estimates

Uncertainty assessment along with forest biomass and carbon reserve estimation are crucial contents required by United Nations Framework Convention on Climate Change(UNFCCC), and regularly reporting them is the obligation of all contracting parties. Based on forest inventory data, Above-ground forest biomass at regional-level is typically estimated by adding model predictions of biomass from individual trees in a plot, and subsequently aggregating predictions from plots to large areas. There are multiple sources of uncertainties throughout this aggregated process. These uncertainties always affect the precision of large area biomass estimates, and the effects are generally overlooked; however, failure to account for these uncertainties will cause erroneously optimistic precision estimates. Therefore, accurately clarifying and assessing the uncertainty caused by all sorts of error sources are significant to imrove the measuring precision of biomass, carbon reserve and corresponding dynamic variation and to provide reliable scientific basis for greenhouse gas emissions carbon sink list editing in our country. Data in this paper were obtained from permanent sample plots in Jiang Xi Province in China including destructively sampled individual trees of Chinese fir(Cunninghamia lanceolata) and Masson pine(Pinus massoniana) from June to September in 2009 for biomass model establishing. Commonly used biomass models and biomass expansion factor were used to estimate regional scale biomass, then Model-depended method, Monte Carlo simulation, Error propagation and Improved Model-depended method were applied to assess uncertainties associated with multiple sources of errors such as measuring error, sampling error and modeling error. Additionally, the influence from model forms, sample size residual variation, error in independent variables were also studied.(1) Three forms of biomass model were applied to estimate the above-ground biomass and tree components(including bole, bark, branch and foliage) of Chinese fir and Masson pine of Jiangxi province, and uncertainties resulted respectively from sampling error and modeling error were assessed by applying Model-depended method. Results showed that using the model forms of bg ?a ?D ??, b cg ?a ?D ?H ?? and 2()bg ?a ?D H ??, the above-ground biomass of Chinese fir in Jiangxi were respectively 19.67±1.37 t/ha, 17.36±1.09 t/ha and 17.07±1.11 t/ha, and those of Masson pine were 20.45±1.40 t/ha, 18.29±1.29 t/ha and 17.20±1.35 t/ha. Estimations under different model forms varied from each other obviously, especially that modeling error played a more significant role in biomass uncertainty than sampling error. Among all tree components, lower uncerainty caused by modeling error was propagated in bole and bark biomass estimation than branch and foliage.(2) Monte Carlo simulation was applied in the above-ground biomass estimation and responding uncertainty assessment of Chinese fir and Masson pine of Jiangxi province, then the impact of residual variation in biomass model to biomass uncertainty was studied. Results revealed that the above-ground biomass of Chinese fir in Jiangxi was 19.67±1.26 t/ha, and the uncertainty accounted for 6.41% of the biomass estimates; the above-ground biomass of Masson pine was 20.50±1.50 t/ha, and the uncertainty accounted for 7.33%. The effect of residual variability associated with R2 was less important in model uncertainty of biomass estimates, however, higher R2 did reduce the operation times for achieving stability of Monte Carlo simulations.(3) Introducing in Monte Carlo simulation, Model-depended method was improved, then the above-ground biomass was estimated and the uncertainties caused by sampling error and modeling error were assessed respectively using this approach. And the influence of different levels of sample size on biomass uncertainty was also studied. Results revealed that using the improved Model-depended method, both the advantage of stability supported by Monte Carlo simulation and of possibility to assess uncertainties affected by sampling error and modeling error separately supported by Model-depended method were achieved. Moreover, with gradually larger modeling sample size, the uncertainty values decreased and the operation times required for achieving the stability of average biomass and corresponding uncertainty in Monte Carlo simulations also were reduced, indicating that increasing modeling sample size is an effective way to increase the precision and accuracy of estimating and reduce uncertainty in regional-level biomass estimations.(4) Error propagation was used to estimate the uncertainty resulted respectively from measuring error, sampling error and modeling error in the biomass estimating of Chinese fir and Masson pine of Jiangxi province. The uncertainty caused by measuring error was under the assumption that measuring error in diameter was follow a normal distribution with zero mean and with standard deviation of 5% of diameter measurement, and that in height was follow a normal distribution with zero mean and with standard deviation of 10% of height measurement. Result revealed that the uncertainty caused by measuring error accounted for 0.08% of the Chinese fir biomass estimates and 0.21% the Masson pine, indicating that the measuing error had negligible impact on regional-scale biomass estimation.(5) Biomass expansion factor(BEF) was used to estimate the regional scale biomass and the uncertainty in BEF was first assessed by applying Monte Carlo simulation, then the sensitivity analysis was conducted to study the effects on BEF from five possible error interference items including error in biomass model, error in timber volume model, error in both biomass model and timber volume model, error in diameter measurement and error in height measurement. Results showed that the biomass expansion factor of Chinese fir biomass was 0.509±0.042 t/m3 and that of Masson pine was 0.630±0.042 t/m3; among five error interference items, error in biomass model, error in timber volume model and error in both biomass model and timber volume model played a more important role in biomass uncertainty in BEF and biomass estimation than error from other sources.