Global warming is currently a major concern to the sustainability of earth′s ecosystems. To mitigate this effect, it is essential to provide policy makers with accurate information on the distribution and dynamics of carbon sources and sinks. However, one important challenge in the estimation of forest carbon is how to quantify its spatial distributions and corresponding uncertainties. This study developed a general methodology for mapping forest carbon sinks by combining existing National Forest Inventory (NFI) plot data and satellite images. This method was based on sequential Gaussian co-simulation to spatially combine and up-scale plot data and satellite images from small (30 m×30 m) to large map units 900 km×900 km. Those large units are usually required for mapping forest carbon at regional, national and global scales. The proposed method was applied to mapping forest carbon using the 2004 NFI plot data and Landsat Thematic Mapper images for Lin′an County, Zhejiang, China. Results showed that the proposed method accurately reproduced the spatial distribution of the NFI plot data. However, the simulated average carbon was 24.9% lower than that of the NFI plot data. This study also provided quantitative information on variability in forest carbon and uncertainty of its estimates, including variances and probability that estimates would be larger than a threshold value. This information will in turn be useful in further uncertainty propagation modeling and analysis, and management of forest carbon markets. In conclusion, this study provided the solution to overcome some of the current significant gaps in the generation and assessment of forest carbon products and its uncertainty.