Surface air temperatures in the Arctic were the second warmest in 2019 since 1900 with widespread consequences for ecosystems and Arctic communities. Physical feedbacks such as the permafrost carbon (C) feedback to climate change are projected to strengthen with each year of warming leading to increased C release to the atmosphere. Capturing permafrost C dynamics in process-based Earth System Models (ESMs) is crucial for accurately projecting the magnitude and timing of permafrost C emissions and thus the rate of global climate change. While numerous experimental and modeling efforts are improving understanding of the magnitude and underlying mechanisms of permafrost C dynamics, there is a disconnect between experimental and modeling approaches and large uncertainties remain.
The complexity of ESMs and the pressing need to reduce model spread requires innovative approaches to evaluate and analyze model results. One promising avenue is a focused effort to constrain model forecasts with experimental data. Manipulative field-based experiments provide an opportunity to inform long-term projections about the magnitude and underlying mechanisms of permafrost C dynamics in a changing Arctic.
Here, we propose to 1) synthesize pan-Arctic experimental warming studies; 2) perform multi-model site-level simulations that align with experimental perturbation approaches to assess coherence among models, evaluate model performance against benchmarks, and inform future measurements; and 3) assess implications of benchmarking for pan-Arctic upscaling and forecasts. The proposed work will be facilitated by a series of virtual and in-person meetings. Community engagement will be an integral aspect of the proposed work as we will leverage the science community to actively participate and contribute data and model resources throughout the project.
Field warming experiments span a range of manipulation types ranging from open-top chambers (primarily air warming), snow fences (primarily soil warming and moisture inputs), and a combination of snow addition and removal (primarily soil warming alone). These experimental approaches can be standardized using quantitative temperature metrics to define the amount of air and soil warming for each experiment. We propose to synthesize data from these experimental manipulations across the pan-Arctic to create new functional benchmarks for evaluating model performance. The benefit of functional benchmarks is that they can provide insight into the potential predictive power of a model, inform poor or missing model representations, and enable the extrapolation of observations beyond sparse study sites.
Using an ensemble of land models, we will perform simulations that align with experimental warming approaches described in the previous section. In addition to evaluating individual model performance, our proposed model intercomparison approach allows us to assess predictive capability among land models, evaluate benchmarks, and develop appropriate metrics for the International Land Model Benchmarking project.
In addition to the site-level runs, we will work with the modeling community to perform pan-Arctic simulations covering time frames spanning the recent historical period (1960-present) and out to the end of the century and beyond (2050-2100, 2300). A key question to be addressed is whether model performance is consistent across scales and whether models that perform well at the site level against synthesized benchmarks, perform equally well at regional scales.