2024 Abstracts

Incorporation of Diverse Arctic Vegetation Types in a Land-Surface Model Improves Representation of Spatial Variability in Carbon Dynamics Across a Tundra Landscape


Bailey A. Murphy1* (murphyba@ornl.gov), Benjamin N. Sulman1, Fengming Yuan1, Verity G. Salmon1, Jitendra Kumar1, Colleen Iversen1, Sigrid Dengel2, Margaret S. Torn2


1Oak Ridge National Laboratory, Oak Ridge, TN; 2Lawrence Berkeley National Laboratory, Berkeley, CA



Earth system models (ESMs), particularly land surface components that simulate complex interactions between vegetation and the atmosphere, serve as essential tools for comprehending climate change. The Arctic is experiencing warming trends surpassing those observed elsewhere on Earth. Given the considerable carbon stored in permafrost that is susceptible to release with ongoing warming, permafrost thaw has the potential to significantly impact global carbon cycling and ecosystem-climate feedbacks. However, model predictions of climate change in Arctic regions are uncertain, in part due to challenges in capturing the complexity of Arctic ecosystems and the historically limited availability of observational data for model validation and calibration. Arctic ecosystems are typically represented in a highly simplified manner. For instance, the Energy Exascale Earth System Model (E3SM) divides Arctic vegetation into two plant functional types (PFTs): deciduous shrub or grass. Recent efforts have integrated diverse Arctic PFTs into the E3SM Land Model (ELM), leveraging the wealth of data from the Next-Generation Ecosystem Experiments (NGEE) Arctic to parameterize the additional PFTs. In this study, researchers conducted spatially explicit 100 × 100 m resolution simulations of vegetation and associated carbon dynamics at the NGEE Arctic Council study site on Alaska’s Seward Peninsula, using ELM configured with the improved representation of Arctic vegetation.

Comparison of model predictions with observational data (e.g., eddy covariance flux tower, airborne hyperspectral remote sensing, and machine learning–based plant community characterization) reveals that the refined PFT representation produced more realistic spatial patterns of tundra vegetation biomass and land-atmosphere exchanges compared to the original model. Variability in vegetation biomass and productivity across the tundra landscape was much higher when incorporating multiple tundra plant growth forms. Results demonstrate that enhanced model representation of the diversity of Arctic vegetation can provide a more robust predictive understanding of carbon dynamics across tundra landscapes, facilitating regional to pan-Arctic scaling.