Representativeness Analysis Identifies Arctic Tundra Areas Poorly Represented by CO2 and CH4 Flux Measurements
Large changes in the Arctic carbon balance are expected as warming linked to climate change threatens to destabilize ancient permafrost carbon stocks. The eddy covariance (EC) method is an established technique to quantify net losses and gains of carbon between the biosphere and atmosphere at high spatiotemporal resolution. Over the past decades, a growing network of terrestrial EC tower sites has been established across the Arctic, but a comprehensive assessment of the network’s representativeness within the heterogeneous Arctic region had not been performed. This study provides an inventory of Arctic EC sites and characterizes the similarity between bioclimatic conditions at sites monitored by the network of EC sites as a means of assessing the representativeness of the monitoring network. To map the representativeness of the EC network, we applied a machine learning method across the pan-Arctic region based on 18 bioclimatic and edaphic variables. Half the Arctic tundra biome is well represented by at least one EC tower, but many EC sites do not measure methane or wintertime greenhouse gas fluxes. Environmental conditions in Fennoscandia and Alaska, where the majority of EC towers are located are well-represented; however, large parts of Siberia and parts of Canada are poorly represented, as are mountainous regions.
We tested three different strategies to identify new site locations or upgrades to existing sites that optimally enhance the representativeness of the current EC network. While 15 new sites can improve the representativeness of the pan-Arctic network by 20%, upgrading as few as 10 existing sites to capture methane fluxes or remain active during wintertime can improve their representative network coverage by 28% to 33%. This targeted network improvement could be shown to be clearly superior to an unguided selection of new sites, therefore leading to substantial improvements in network coverage based on relatively small investments.
We applied a machine learning method based on 18 bioclimatic and edaphic variables to map the representativeness of a network of eddy covariance (EC) in the pan-Arctic region. This study provides an inventory of Arctic EC sites and characterizes the similarity between bioclimatic conditions at sites monitored by the network of EC sites as a means of assessing the representativeness of the monitoring network. Half of the Arctic tundra biome is well represented by at least one EC tower, but many EC sites do not measure methane or wintertime greenhouse gas fluxes. While 15 new sites can improve the measurement coverage of the network by 20%, upgrading as few as 10 existing sites to capture methane fluxes or remain active during wintertime can improve network representativeness by 28% to 33%. This analysis advances the understanding of measurement gaps and informs future investments in infrastructure required to improve understanding of the rapidly changing Arctic.