Comparison of the CFLOW and CARBINE carbon accounting models

have been applied extensively to the quantification of forest carbon stocks and potential stock changes across a range of scales from stand to national levels. The principal objectives of this study are an evaluation of the completeness and robustness of the C-FLOW and CARBINE models for estimating carbon stocks and potential stock changes in the forestry sector at the stand and national levels, and the reliability of underpinning data and parameter estimates used by C-FLOW and CARBINE. Both models use a very similar methodology to estimate total tree carbon stocks. It is difficult to assess which model estimates total forest carbon more accurately as very little information was provided on the scientific basis for the expansion factors to relate stem carbon to other forest components. C-FLOW explicitly includes litter in the model while in CARBINE it is not clear if, where or how litter is included. Each model includes a soil carbon sub-model but the methodology used to estimate soil carbon is quite different. In C-FLOW this is linked to the litter sub-model and soil carbon changes are linked to litter inputs and decay, while in CARBINE the soil sub-model is run completely independently and soil carbon change is based on land-use change. The two models CARBINE and C-FLOW usually predict very similar results (within 12% of each other) in terms of carbon stocks in trees. The accuracy of predictions of carbon stocks in forest biomass produced by C-FLOW and CARBINE was first assessed by comparison with the independently-developed BSORT model. Both models were observed to make predictions that were systematically inconsistent with those of BSORT, although differences were less significant for conifer species than for broadleaf species. The accuracy of predictions of forest biomass carbon stocks made by C-FLOW and CARBINE was further assessed by comparison with results derived from site-specific permanent mensuration sample plot data. Although this analysis can only be regarded as an initial investigation, the results indicated that the accuracy of predictions made by both models is well within short-term fluctuations observed for individual stands (±10%). There may be further, significant issues of accuracy arising from the limited range of combinations of species, yield class and management regime covered by both models. Both tree species and yield class can be expected to have a significant influence on the timecourse of accumulation of forest biomass carbon stocks. Most significant of all is likely to be management regime. …