OOS 13-8 - Importance of local Arctic tundra heterogeneity for regional carbon modelling

Wednesday, August 14, 2019: 10:30 AM
M103, Kentucky International Convention Center
Mark J. Lara1,2,3, A. David McGuire3, Eugenie Euskirchen3, Hélène Genet3 and Stan D. Wullschleger4, (1)Department of Plant Biology, University of Illinois, Urbana, IL, (2)Department of Geography, University of Illinois, Urbana, IL, (3)Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, AK, (4)Environmental Sciences Division and Climate Change Science Institute, Oak Ridge National Laboratory, Oak Ridge, TN
Background/Question/Methods

There are considerable uncertainties regarding the fate of the extensive soil carbon pool with projected climate warming over the next century. In northern Alaska, nearly 65% of the terrestrial surface is composed of polygonal tundra, where geomorphic landforms disproportionately influence local surface hydrology and soil carbon and nutrient cycling over small spatial scales. Process-based biogeochemical models used for local to global projections of ecological responses to climate change, typically operate at larger scales (1 km2 to 0.5°) at which fine scale (<1 km) spatial heterogeneity is often aggregated to the dominant land cover unit. Here we evaluate the importance of Arctic polygonal tundra heterogeneity for representing soil carbon dynamics at fine (30 m) to coarse (~0.5°) spatial scales and with abundant (6 landforms) to limited (1 landform) model parameterization data. We leverage the legacy of data collected near Utqiaġvik (formerly Barrow) Alaska between 1973 to 2016 to represent six of the most dominant polygonal tundra landforms, ranging from dry high-center polygons to aquatic polygonal tundra ponds.

Results/Conclusions

Model simulations spanning the Barrow Peninsula (~2000 km2 ) identify error to linearly increase as spatial resolution and the number of modelled land cover classes decrease. This work suggests representing Arctic tundra heterogeneity matters, however we provide a simple solution to minimize model simulation error in large-scale assessments.