Understanding the representativeness of FLUXNET for upscaling carbon flux from eddy covariance measurements

Authors

Forrest M. Hoffman (forrest at climatemodeling dot org)
Oak Ridge National Laboratory
Jitendra Kumar
Oak Ridge National Laboratory
William Walter Hargrove
USDA Forest Service Southern Research Station
Nathaniel Collier
Oak Ridge National Laboratory

Session

S5. Applications of macrosystems ecology for climate and landscape change
Monday, April 10, 2017 3:20 pm–3:40 pm
Constellation E

Abstract

Eddy covariance data from regional flux networks are direct in situ measurements of carbon, water, and energy fluxes and are of vital importance for understanding the spatio-temporal dynamics of the the global carbon cycle. FLUXNET links regional networks of eddy covariance sites across the globe to quantify the spatial and temporal variability of fluxes at regional to global scales and to detect emergent ecosystem properties. We performed an assessment of the global representativeness of FLUXNET based on the recently released FLUXNET2015 data set. A detailed high resolution analysis of the evolving representativeness of FLUXNET through time will be presented. Results provide quantitative insights into the extent that various biomes are sampled by the global network of networks, the role of the spatial distribution of the sites on the network scale representativeness at any given time, and how that representativeness has changed through time due to changing operational status and data availability at sites in the network. To realize the full potential of FLUXNET observations for understanding emergent ecosystem properties at regional and global scales, we developed an approach for upscaling eddy covariance measurements. Informed by the representativeness of observations at the flux sites in the network, the upscaled data reflects the spatio-temporal dynamics of the carbon cycle captured by the in situ measurements. This study provides a method for optimal use of the rich site measurements from FLUXNET to derive an understanding of upscaled carbon fluxes, which can be routinely updated as new data become available, and direct network expansion by identifying regions poorly sampled by the current network.


Forrest M. Hoffman (forrest at climatemodeling dot org)