of the major U.S. ecosystem and climate types (Hargrove et al. 2003; Xiao et al.
2011a), some geographical regions and vegetation types are still underrepresented.
For example, there are very few sites in the Great Basin, the Rocky Mountain, and
the western Great Plains regions. There are also very limited sites for open
shrublands and savannas. The temporal representativeness of the AmeirFlux data
may also influence the flux estimates. Multiple years of data (2000–2004) were
used to train the models here to account for the interannual variability of fluxes.
A 5-year period of time, however, is perhaps still limited for capturing some
extreme events.
6.6 Future Research Directions
Satellite remote sensing provides valuable information for upscaling flux observations from the tower footprint to regional and continental scales. The resulting
gridded flux estimates generally capture the spatial and temporal patterns of NEE.
These flux estimates can be used to examine the magnitude, distribution, and
interannual variability of net carbon uptake/release over broad regions.
In future work, the upscaling of flux observations should explicitly incorporate
the impacts of disturbance on ecosystem carbon exchange (e.g., Amiro et al. 2010;
Liu et al. 2011). Satellite remote sensing can be used to produce spatially-explicit
information on aboveground biomass (Zhang and Kondragunta 2006), disturbance
(Goward et al. 2008; Huang et al. 2010), and stand age (Pan et al. 2011b), which is
potentially useful for accounting for the state and stages of forest ecosystems and
the impacts of disturbances. Future upscaling work is also expected to advance
towards quantifying uncertainties associated with gridded flux estimates by considering various sources of uncertainty (Xiao et al. 2011b, 2012). Future upscaling
efforts will benefit from the intercomparison of multiple upscaling methods
including data-driven (e.g., Xiao et al. 2008, 2010, 2011a; Jung et al. 2009; Zhang
et al. 2011) and data assimilation (e.g., Xiao et al. 2011b) approaches and the
resulting flux fields. The intercomparison of flux estimates resulting from different
upscaling approaches as well as comparison of these approaches to other methods
such as atmospheric inversions, biomass inventories, and ecosystem models can
provide complementary information for the diagnostics of net carbon exchange
between the terrestrial biosphere and the atmosphere and valuable information for
future improvement of these approaches (Xiao et al. 2012).
Acknowledgments This work is supported by National Science Foundation (NSF) through
Macrosystems Biology program under award 1065777, National Aeronautics and Space
Administration (NASA) through Carbon Monitoring System (CMS) under grant NNX11AL32G,
and Department of Energy (DOE) through National Institute for Climatic Change Research
(NICCR) under grant 14U776. I thank the research/technical personnel of the AmeriFlux towers,
MODIS data products, and MERRA data products for making the flux observations, MODIS data
streams, and MERRA data available, respectively. I also thank the two anonymous reviewers for
their constructive comments on the manuscript.
164
J. Xiao
2011a), some geographical regions and vegetation types are still underrepresented.
For example, there are very few sites in the Great Basin, the Rocky Mountain, and
the western Great Plains regions. There are also very limited sites for open
shrublands and savannas. The temporal representativeness of the AmeirFlux data
may also influence the flux estimates. Multiple years of data (2000–2004) were
used to train the models here to account for the interannual variability of fluxes.
A 5-year period of time, however, is perhaps still limited for capturing some
extreme events.
6.6 Future Research Directions
Satellite remote sensing provides valuable information for upscaling flux observations from the tower footprint to regional and continental scales. The resulting
gridded flux estimates generally capture the spatial and temporal patterns of NEE.
These flux estimates can be used to examine the magnitude, distribution, and
interannual variability of net carbon uptake/release over broad regions.
In future work, the upscaling of flux observations should explicitly incorporate
the impacts of disturbance on ecosystem carbon exchange (e.g., Amiro et al. 2010;
Liu et al. 2011). Satellite remote sensing can be used to produce spatially-explicit
information on aboveground biomass (Zhang and Kondragunta 2006), disturbance
(Goward et al. 2008; Huang et al. 2010), and stand age (Pan et al. 2011b), which is
potentially useful for accounting for the state and stages of forest ecosystems and
the impacts of disturbances. Future upscaling work is also expected to advance
towards quantifying uncertainties associated with gridded flux estimates by considering various sources of uncertainty (Xiao et al. 2011b, 2012). Future upscaling
efforts will benefit from the intercomparison of multiple upscaling methods
including data-driven (e.g., Xiao et al. 2008, 2010, 2011a; Jung et al. 2009; Zhang
et al. 2011) and data assimilation (e.g., Xiao et al. 2011b) approaches and the
resulting flux fields. The intercomparison of flux estimates resulting from different
upscaling approaches as well as comparison of these approaches to other methods
such as atmospheric inversions, biomass inventories, and ecosystem models can
provide complementary information for the diagnostics of net carbon exchange
between the terrestrial biosphere and the atmosphere and valuable information for
future improvement of these approaches (Xiao et al. 2012).
Acknowledgments This work is supported by National Science Foundation (NSF) through
Macrosystems Biology program under award 1065777, National Aeronautics and Space
Administration (NASA) through Carbon Monitoring System (CMS) under grant NNX11AL32G,
and Department of Energy (DOE) through National Institute for Climatic Change Research
(NICCR) under grant 14U776. I thank the research/technical personnel of the AmeriFlux towers,
MODIS data products, and MERRA data products for making the flux observations, MODIS data
streams, and MERRA data available, respectively. I also thank the two anonymous reviewers for
their constructive comments on the manuscript.
164
J. Xiao