Possible effects of altered growth behaviour of Norway spruce (Picea abies) on carbon accounting

Implementing the Kyoto Protocol necessitates precise and practical inventory methods for estimating the carbon reservoir in forests. An analysis of the German state North Rhine-Westphalia forest inventory data showed enormous deviations from the yield tables of Wiedemann (1936, 1942). Specifically the ratio of height/diameter at breast height was up to 30% lower than in the yield tables, which indicated wider annual rings and a lower basic density. Such differences most likely pose consequences for the calculation of the amount of carbon in these forests. In this study the aboveground green weight, moisture and the carbon content of fifteen 46-year old Norway spruces (Picea abies) were measured, and two approaches for calculating the aboveground carbon content of trees were discussed: the common methodology of applying expansion and conversion factors as well as using optimized regression equations. The average expansion factor deviated approximately 1.2% from the yield tables of Grundner and Schwappach (1906). The wood density was 368 kg/m(3) and within the European scope of 320-420 kg/m(3) (Bosshard, 1984; Hakkila, 1989; Knigge and Schulz, 1966; Trendelenburg and Mayer-Wegelin, 1955). The woody carbon content was found to be 50.42%, which corresponded well with reference values of 50.3% (Bosshard, 1984) and 51.4% (Knigge and Schulz, 1966). However, the average percentage of crown material carbon content was 0.75% higher than in the tree boles. This study demonstrated that the most reliable methodology for calculating aboveground tree carbon was a power regression function, which calculated the growing stock volume, converting it directly into tree carbon content. The results deviated by less than 1.3% (R-2 = 0.99) from the measured tree carbon. Using volume expansion factors and factors of basic density and carbon percentage, as in Burschel et al. (1993), the calculated results diverged -3.7% from the measured outcome. The adapted form according to Fang et al. (2001) deviated -1.2% (R-2 = 0.63) and the form following Brown and Schroeder (1999) about -2.5% (R-2 = 0.61) from the measured carbon of the sample trees. However, coefficients of variation from 12% and 13% for the presented algorithm illustrated that in reality the calculated carbon of single trees could diverge considerably from the model results.