<p>Northern peatlands (>45°N) mostly initiated during the Holocene and have been a large C sink to the atmosphere. Northern peatland formation prefers wet and cold condition where the productivity persistently exceeds decomposition and thereby C accumulates. As the northern high latitude region is likely to be warmer in the future, whether northern peatlands will continue being C sinks or switch to C sources is uncertain. To address this issue, I revise and apply a process-based model designed for describing peatland biogeochemical processes, Peatland Terrestrial Ecosystem Model (PTEM), to simulate the C dynamics at both site and regional level, from 15 ka BP-2300. For the site-level simulation, PTEM 1.0 is substantially revised into PTEM 2.0 in terms of peat accumulation process, plant functional types, productivity and decomposition, and soil thermal properties. A simulation from peat initiation to 2300 is conducted for three northern peatland sites. I found PTEM 2.0 can effectively capture the historical C accumulation progress, when compared with the observation. The future simulation indicates northern peatlands have reduced C sink capacity or switch to a C source under N insufficiency and water table deepening. </p> <p>Afterwards, a historical pan-Arctic simulation during 15ka BP-1990 is conducted. PTEM 2.0 is revised into PTEM 2.1 by adding spatially-explicit run-on and run off processes. The spatially-explicit peat initiation dataset is derived from neural network approach and a spatially-explicit peat expansion trend is established on top of it. My estimated pan-Arctic peatland C storage is 396-421 Pg C with the long-term C accumulation rate (CAR) of 22.9 g C∙m-2 yr-1. The simulated spatial distribution of peat C and the temporal pattern of CAR both agree with literature values. I analyzed northern peatlands’ response to historical climate change since 0.5 ka BP and found decreased CAR in the warmer non-permafrost and permafrost-thaw region, while the opposite was found in the colder permafrost region. The results indicate warmer southern peatlands will first switch to a C source under warming while more northern peatlands will become larger sinks. </p> <p>Based on the result of historical simulation, a future simulation is conducted for 1990-2300 with peatland expansion/shrinkage considered. PTEM 2.1 is revised into PTEM 2.2 such that the water table depth, run-on and run-off are estimated from a TOPMODEL approach. In the 21st century, northern peatlands are projected to be a C source of 1.2-13.3 Pg C under five out of six climate scenarios. During 2100-2300, northern peatlands under all scenarios are a C source under all climate scenarios. Northern peatlands switch to C sources due to deepening water table depth, insufficient N availability, and plant functional type shift. I found that northern peatlands remain as a C sink until a mean pan-Arctic peatlands annual temperature reaches -2.09 - -2.89°C. This study predicts a northern peatland sink to a source shift around 2050, earlier than previous estimates of after 2100, and emphasizes the vulnerability of northern peatlands to climate change. </p>