Mutli-source approach to forest carbon inventories
The Kyoto Protocol to the United Nations Framework Convention on Climate Change set targets for limiting greenhouse gas emissions and required the implementation of national systems for their monitoring. Currently, greenhouse gas emissions and removals related to forestry activities are reported using information from National Forest Inventories, which focus only on stem wood. Improving the completeness of greenhouse gas reporting at national level, as well as at the level of the European Union, was the overall aim of the CarbonInvent project. Data from multiple sources such as soil carbon observations, available inventory data and remote sensing imagery were integrated. Additionally, new links were established between National Forest Inventories and greenhouse gas reporting needs through improved and consistent methods for reliable estimates of carbon stocks in European forests. Traditional forest inventories typically provide information on stem volumes, but not on biomass or carbon stock. Thus available tree volume estimates had to be converted into biomass and carbon stock estimates by means of straightforward biomass expansion factors (BEFs) or biomass functions. The latter utilize individual tree measurements, mostly diameter or tree height, as input variables in order to predict age- and species- specific tree biomass. The individual methodological components were studied in test sites representing the major forest regions in Europe (Catalonia in Spain, Salzburg in Austria, Thuringa in Germany and Hyytiala in Finland). The diverging results of the two approaches showed the potential of the estimation methods in terms of assessment accuracy and sampling error propagation. A multiple source inventory concept that utilizes field measurements combined with auxiliary data sources proved to enhance accuracy in terms of sampling errors and thus resulted in more reliable methods for carbon monitoring in forest areas. However, BEFs and biomass functions were developed for local applications and specific tree species and as a consequence fail to provide accurate estimates for a wide variety of site conditions. Further improvements are absolutely mandatory to increase the method's feasibility for operational applications.
