North American Net Ecosystem Exchange: Regionalization of AmeriFlux Observations

Anthony W. King, W. M. (Mac) Post, Stan D. Wullschleger, Holly K. Gibbs

Environmental Sciences Division, Oak Ridge National Laboratory,

Oak Ridge, TN 37831-6335

INTRODUCTION: Observations of net ecosystem exchange (NEE, g C m-2y-1) from the AmeriFlux network of eddy-covariance systems (Figure 1) provides a basis for estimating NEE of North America, with implications for assessing the magnitude and nature of a North American carbon sink. The challenge lies in the application of appropriate techniques for scaling-up point estimates of NEE from AmeriFlux sites to regional estimates. This use of the AmeriFlux data complements the use of the AmeriFlux data to understand NEE processes. We are exploring a sequence of increasingly refined, and more difficult, methods for regionalization of NEE from AmeriFlux site data. These methods range from simple methods based on defining the geographical areas that sites represent and multiplying site estimates by those areas, to more refined methods involving the application of geographically distributed, process-based terrestrial ecosystem models. Here we present first approximations of NEE for the forested region of the eastern United Sates using the simplest of these methods, extrapolating mean values over appropriate areas. We discuss the refinements needed to improve these estimates and point towards the next steps of more sophisticated methods of regionalization.

Figure 1. The AmeriFlux network of eddy-covariance sites in the United States and Canada.

Table 1

Table 1. AmeriFlux eddy-covariance sites with published annual NEE for forests of the eastern United States. Forest types are those given in the AmeriFlux site descriptions.

Site

1992

1993

1994

1995

1996

1997

1998

1999

2000

Reference

Duke Forest







-538

-651


6

Florida Cypress





-84

-37




2

Florida Slash Pine





-740

-610




2

Harvard Forest

-200

-190

-200

-250

-200

-210

-120

-230

-210

1

Howland Forest





-246

-256




5

Morgan Monroe







-240



8

Park Falls






-34

55

44


6

U. Mich. Bio. Stat.








-167


3

Walker Branch



-525

-470

-576

-618

-870



10

Willow Creek








-180


6

Table 2. Published estimates of annual NEE (g C m-2y-1) from AmeriFlux sites in forests of the eastern United States. Negative values indicate net ecosystem uptake and storage.

METHOD: Simple Area Extrapolations.We have identified 10 AmeriFlux sites in the forested region of the eastern United States (Figure 2) with published estimates of annual NEE (Tables 1 and 2). The areas represented by the AmeriFlux sites were determined by referencing the sites and their geographical coordinates against satellite-derived data bases describing the contemporary vegetation/land cover of the United States. We used the National Land Cover Data map produced from 30m Landsat scenes collected in 1992 (Vogelman et al. 2001), the University of Maryland map produced from 1km Advanced Very High Resolution Radiometer (AVHRR) data collected in 1992-93 (Hansen et al. 2000), and the U.S. Forest Service map produced from 1km AVHRR data collected in 1992 (Powell et al. 1992). Estimates of NEE from multiple sites occurring in different forest types, or aggregated types, were averaged, and these averages scaled to the forest type and region by multiplying the mean flux value by the area of the type.

Figure 2. Distribution of AmeriFlux sites in the forests of the eastern United States with published estimates of annual NEE. Mapped forest types are those of the U.S. Forest Service.

RESULTS: The occurrence of the AmeriFlux sites in different eastern United States forest types based on site coordinates is shown in Table 3. The National Land Cover Data is mapped at 30m resolution, the other data are both mapped at 1km; some deviation from the vegetation type described for the tower (Table 1) is expected with coarser resolution in heterogeneous landscapes. A translation from the vegetation described for the sites (Table 1) and the mapped vegetation types has also been produced but is not shown here.

As an initial approximation, the annual NEE values (Table 2) are first averaged across years, providing a characteristic, time-averaged NEE for each site. (For some sites only a single annual value is available). The sites are then binned by mapped forest types (Table 3) and a spatial average of the time-averaged, site-specific NEE is calculated for each forest type or aggregate of those types. The spatially-averaged NEE (g C m-2y-1) is scaled to a regional type-specific flux (Tg C y-1) by the area of the type. The sum of the type-specific fluxes is an estimate of net annual change in carbon storage (source or sink) for the forests of the eastern United States. The resulting spatially-averaged NEE and regional fluxes are shown in Table 4 below. Our simple extrapolation of AmeriFlux NEE data suggests that the forests of the eastern United States are currently an appreciable sink for atmospheric CO2 on the order of 0.4-0.6 Gt C y-1. These estimates are consistent with published estimates of carbon uptake by North American forests of approximately 0.5 Gt C y-1 obtained by a variety of alternative methods.

Table 3

Table 3.The distribution of AmeriFlux sites in forests of the eastern U.S. based on flux-tower coordinates (Table 1) for three mappings of U.S. vegetation/landcover. Types in parentheses were deemed more appropriate than the vegetation specified by the actual coordinates. The misclassification may arise from lack of precision in the tower coordinates, error in the vegetation map, or in the nature of the site itself. The corrected types were used in the area extrapolations. (*There are wetlands in the vicintiy of Howland Forest tower. A small error in tower coordinates coul be responsible for the misclasification. ** The Park Falls tall tower stands in a large forest clearing.)

The relative contribution of evergreen versus deciduous forests to the eastern forest sink varies with the forest cover data and with the assignment of AmeriFlux sites to different forest types. In general, however, deciduous forests are the larger contributor because of their larger area.

 

National Land Cover Data

University of Maryland Map

U.S. Forest Service Map

Forest type

Area

(km2)

Average NEE

(g C m-2y-1)

Regional Flux

(Tg C y-1)

Area

(km2)

Average NEE

(g C m-2y-1)

Regional Flux

(Tg C y-1)

Area

(km2)

Average NEE

(g C m-2y-1)

Regional Flux

(Tg C y-1)

Coord. Based

Veg. Based

Coord. Based

Veg. Based

Coord. based

Veg. Based

Coord. Based

Veg. Based

Coord. Based

Veg. Based

Coord. Based

Veg. Based

Evergreen

402373

-507130

-20152

-20452

-20452

232093

-368307

-507130

-8571

-11830

419416

-2510

-507130

-1050

-21355

Deciduous

1173246

-20682

-259104

-24296

-304122

526874

-253102

-259104

-13354

-13655

890552

-30186

-259104

-26877

-23193

Mixed

351100

0

-7394

0

-2633

408837

-349124

-7394

-14351

-3038

168062

0

-7394

0

-1216

Woodland

0

0

0

0

0

537500

-1670

0

-900

0

0

0

0

0

0

Total

1926719

-29677

-29677

-570148

-570148

1704854

-29677

-29677

-505131

-505131

1478030

-29677

-29677

-437114

-437114

Table 4. Mean NEE and regionalized carbon flux for the major forest types of the eastern United States NEE and regional flux values are reported as the mean standard error. Negative values represent net ecosystem uptake and a regional sink for atmospheric CO2. Flux tower coordinates define the forest type (Table 3) for the coord.-based estimates (Table 3); site descriptions of vegetation define the forest type for the veg.-based estimates (data not shown).

CONCLUSIONS AND FUTURE DIRECTIONS: The results presented here are simply a first step in the regionalization of AmeriFlux NEE data to estimate carbon uptake and storage by the vegetation of North America. A very simple spatial extrapolation provides reasonable results for carbon sequestration by forests of the eastern United States (0.4-0.6 Gt C y-1). These results are promising and suggest that refinements of even this simple extrapolation such as characterization of ecosystem types and areas by both soils and vegetation rather than vegetation alone can provide even more robust results. We are pursuing more sophisticated methods of regionalization, including the use of empirical models of spatial and temporal variability in NEE observed at AmeriFlux sites to define spatially distributed data critical for regionalization (see References 1 and 6). We are also employing a geographically distributed, process-based model (GTEC 2.0) at high spatial resolution (1km) across North America with model NEE calibrated and constrained at AmeriFlux sites. But these more sophisticated techniques are also more data intensive and less analytically tractable. Obtaining good results will take time. In the interim, simple extrapolations like those presented here can provide good first approximations of carbon uptake and release by terrestrial ecosystems of the United States and North America.

REFERNCES

1. Barford, C.C. et al. 2001. Factors controlling long-and short-term sequestration of atmospheric CO2in a mid-latitude forest. Science 294:1688-1691.

2. Clark, K.L., H.L. Gholz, J.B. Moncrieff, F. Cropley and H.W. Loescher. 1999. Environmental controls over net exchanges of carbon dioxide from contrasting Florida ecosystems. Ecological Applications 9:936-948.

3. Curtis, P.S., P.J. Hanson, P. Bolstad, C. Barford, J.C. Randolph, H.P. Schmid, and K.B. Wilson. 2002.Biometric and eddy-covariance based estimates of annual carbon storage in five eastern North American deciduous forests.Agricultural and Forest Meteorology (in press).

4. Hansen, M., DeFries, R. Townshend, J. R. G. and Sohlberg, R. 2000. Global land cover classfication at 1km resolution using a decision tree classifier. International Journal of Remote Sensing. 21: 1331-1365.

5. Hollinger, D.Y., S. M. Goltz, E.A. Davidson, J.T. Lee, K. Tu and H.T. Valentine. 1999. Seasonal patterns and environmental control of carbon dioxide and water vapor exchange in an ecotonal boreal forest. Global Change Biology 5:891-902.

6. Law, B.E., E. Falge, L. Gu et al. 2002. Environmental controls over carbon dioxide and water vapor exchange of terrestrial vegetation.Agricultural and Forest Meteorology (in press).

7. Powell, D.S., J.L. Faulkner, D. Darr, Z. Zhu; D.W. MacCleery. 1992. Forest resources of the United States, 1992. Gen. Tech. Report. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Forest and Range experiment Station. Report is vailable on-line at http://www.srsfia.usfs.msstate.edu/rpa/rpa93.htm.

8. Schmid, H.P., C.S.B Grimmond, F. Crope, B Offerele and H.B. Su. 2000. Measurements of CO2 and energy fluxes over a mixed hardwood forest in the mid-western United States.Agricultural and Forest Meteorology 103:357-374.

9. Vogelmann, J.E., S.M. Howard, L. Yang, C.R. Larson, B.K. Wylie, N. VanDriel. 2001. Completion of the 1990s National Land Cover Dataset for the coterminous United States from Landsat Thematic Mapper data and ancillary data sources. Photogrammetric Engineering and Remote Sensing 67(6): 650 - 661.

10. Wilson, K.B. and D.D. Baldocchi. 2002. Comparing independent estimates of carbon dioxide exchange over 5 years at a deciduous forest in the southeastern United States. Journal of Geophysical Research (in press).

Acknowledgement: This research is supported by the Department of Energys Office of Science Biological and Environmental Research Program.