Biosphere/atmosphere CO₂ exchange in tundra ecosystems: Community characteristics and relationships with multispectral surface reflectance

The spatial and temporal patterns of many of the factors controlling CO 2 exchange are related to characteristics of the vegetated surface which can potentially be monitored using multispectral remote sensors. Realization of this potential depends, in part, on an improved understanding of ecosystem processes and their relationship to variables which are accessible to remote sensing techniques. We examined these relationships using portable, climate‐controlled, instrumented enclosures to measure CO 2 exchange rates in selected tundra sites near Bethel, Alaska. Rates were related to vegetation community type and climatic variables. Exchange rates in enclosures were compared to exchange measurements obtained by eddy correlation on a 12‐m micrometeorological tower. For an average light input of 37 einsteins/day during 20 midsummer days, the empirically modeled exchange rate for a representative area of vegetated tundra was 1.2 ±1.1 (95% confidence interval) g CO 2 m −2 d −1 . This was comparable to a tower measured exchange over the same time period of 1.1 ±1.1 (95% confidence interval) g CO 2 m −2 d −1 . Net exchange in response to varying light levels was compared for two major community types, wet meadow and dry upland tundra, and to the net exchange measured by the micrometeorological tower technique. Portable radiometers were used to measure the multispectral reflectance properties of the sites. These properties were then related to exchange rates with the goal of providing a quantitative foundation for the use of satellite remote sensing to monitor biosphere/atmosphere CO 2 exchange in the tundra biome. The Normalized Difference Vegetation Index (NDVI) and the near‐infrared/red reflectance ratio (SR) computed from surface reflectance were strongly correlated with net CO 2 exchange for both upland and wet meadow vegetation. However, the form of the relationship was distinct from measured correlations in other ecosystems, suggesting that global surveys may require adjustment for geographical differences in exchange processes.