High spatial resolution remote sensing models for landscape-scale CO？ exchange in the Canadian Arctic

ABSTRACT Climate warming is affecting terrestrial ecosystems in the Canadian Arctic, potentially altering the carbon balance of the landscape and contributing additional CO2 to the atmosphere. High spatial resolution remote sensing data can enhance models of net ecosystem exchange (NEE) and its component fluxes, gross ecosystem exchange (GEE), and ecosystem respiration (ER) by quantifying vegetation structure and function over time. In this study, we explored the variability of daytime CO2 exchange rates for three vegetation types along a natural moisture gradient at ecologically distinct mid- and high Arctic sites. We demonstrated that for the two sites studied, there was no statistically significant variation in CO2 exchange rates for the vegetation types through the peak growing season. Hence, the capacity to model these rates with a limited number of satellite data acquisitions is feasible. Simple bivariate models relating the Normalized Difference Vegetation Index (NDVI) to CO2 exchange processes (GEE, ER, and NEE) were developed independent of vegetation type and geographic location and validated using independent data. The spectral models explain between 33 and 94 percent of the variation in CO2 exchange rates at each site, indicating a high level of functional convergence in ecosystem-level structure and function within Arctic landscapes