Multi-model intercomparison of northern peatland carbon cycle

Peatlands cover only 3% of Earth’s land surface but contain about 30% of the global soil carbon pool. Derived predominantly from plant litter and moss accretions, peat deposits are critically sensitive to environmental dynamics such as soil temperature and moisture levels. This vulnerability has raised concerns about potential positive feedback mechanisms in relation to global climate change. However, current global models present disparities in projected emissions, and sensitivities of peatland carbon stocks to changing environments are a major uncertainty in global carbon projections. The Spruce and Peatland Responses Under Changing Environments (SPRUCE) experiment is a large‐scale climate change manipulation that focuses on the combined response of multiple levels of warming at both ambient and elevated CO₂ concentration (aCO₂ and eCO₂), which makes it is a valuable testbed for the broader modeling community to improve the diagnosis and attribution of C fluxes in peatland ecosystems. The SPRUCE Multi-model Intercomparison Project (SPRUCE-MIP) aims to evaluate the projections of peatland carbon cycle dynamics and their warming responses of various models by comparing the model outputs to empirical data from SPRUCE. We assessed 12 different models, focusing on their predictions for net ecosystem carbon exchange (NEE) and its components – net primary productivity (NPP), heterotrophic respiration (HR) and methane (CH₄) fluxes. These predictions were compared to an extensive on-site carbon cycle dataset across five distinct temperature warming levels and two CO₂ concentration scenarios. Our findings revealed significant variability in the model projections, with substantial scatter in the absolute values, warming sensitivities and eCO₂ effects. For example, the models’ prediction of carbon take up between 10 and 732 gC m^-2 year^-1 forthe baseline warming with aCO₂ condition, and the warming sensitivity response is about 1 to 126 gC m^-2 year^-1 ℃^-1. In addition, notable increases productivity under eCO₂ condition are observed in the ORCHIDEE (138.9%), VISIT (105.2%) and ELM-Microbe (64.4%) models while there is no eCO₂ effects for model CoupModel. Experimental measurements showed carbon source even for the baseline warming chamber while most of the model predicted carbon sink, except for model CoupModel, MWM, and PTEM. Furthermore, both the observations and these three models show a significant in C release to atmosphere, making stronger C sources at the extreme +9°C warming level for both aCO₂ and eCO₂ conditions, and models such as CoLM, ELM-SPRUCE and JULES transition from a C sink to a C source under these conditions. Meanwhile, the DNDC, ORCHIDEE and VIST models switch from a C sink to a C source under aCO₂conditions but remain C sinks under eCO₂ conditions. In contrast, models like ELM-Microbe, LPX-Bern and TECO predict that the SPRUCE peatland ecosystem continues to function as a C sink even at the +9°C warming level under both CO₂ conditions. Integrating models with experimental design will allow targeting of these uncertainties and help to reconcile divergence among models to produce more confident projections of peatland ecosystem responses to global changes.