Soil CO2 balance and its uncertainty in forestry-drained peatlands in Finland

To understand the global carbon cycle and the impact of human activity on climate, it is necessary to quantify the net CO2 exchange of ecosystems in different land uses on a large scale. Methods to estimate soil net CO2 exchange (NECO2soil) for drained peatland forests have been largely based on chamber measurements and statistical models. The uncertainty in these methods has not been assessed. Yet, disturbed organic soils are a globally important, potential CO2 source due to their vast carbon storage and its sensitivity to changes in soil moisture. In this study, we estimated the countrywide NECO2soil for the 4.76 million ha of forestry-drained peat soils in Finland. We gathered available litter production and CO2 efflux data and constructed models to be used for the upscaling of NECO2soil from forest inventory data. The contribution of each model and the inventory sampling to the precision of the countrywide estimate was calculated. Also, the sensitivity to possible bias in selected model components was estimated. Compared to the estimated mean NECO2soil, ranging from a source of +20 g m(-2) year(-1) of C to a sink of -40 g m(-2) year(-1) of C, the uncertainty was high. The precision of the estimate (+/- 1 standard deviation) was +/- 20 g m(-2) year(-1) of C. Due to possible bias in the estimated belowground litter input, the overall uncertainty was much higher, around 60 g m(-2) year(-1) of C. The main reason for the high relative uncertainty was NECO2soil being on average close to zero in these boreal forestry-drained peatlands. Forest inventory sample size was large enough and the data for the models were mainly sufficient. To reduce the uncertainty, better understanding of belowground carbon fluxes in order to accurately determine the C input to soil, is crucial. (C) 2014 Elsevier B.V. All rights reserved.