Scaling up ecosystem processes from plots to landscapes is essential for understanding landscape structure and functioning as well as for assessing ecological impacts of land use and climate change. This study illustrates an upscaling approach to studying the spatiotemporal pattern of ecosystem processes in the Changbai Mountain Nature Reserve in northeastern China by integrating simulation modeling, GIS, remote sensing data, and field-based observations. The ecosystem model incorporated processes of energy transfer, plant physiology, carbon dynamics, and water cycling. Using a direct extrapolation scheme, the patch-level ecosystem model was scaled up to quantify the landscape-level pattern of primary productivity and the carbon source-sink relationship. The simulated net primary productivity (NPP) for the entire landscape, consisting of several ecosystem types, was 0.680 kg C m-2 yr-1. The most widely distributed ecosystem type in this region was the mixed broad-leaved and Korean pine (Pinus koraiensis) forest, which had the highest NPP (1.084 kg C m-2 yr-1). The total annual NPP for all ecosystem types combined was estimated to be 1.332 Mt C yr-1. These results suggest that the Changbai Mountain landscape as a whole was a carbon sink, with a net carbon sequestration rate of about 0.884 Mt C yr-1 for the study period. The simulated NPP agreed reasonably well with available field measurements at a number of locations within the study landscape. Our study provides new insight into the relationship between landscape pattern and ecosystem processes, and useful information for improving management practices in the Changbai Mountain Nature Reserve, which is one of the most important forested landscapes in China. Several research needs are discussed to further refine the modeling approach and reduce prediction uncertainties.