Historical climatic variations and rising atmospheric CO₂ concentrations which favor net primary production (NPP) over heterotrophic respiration (R_H) may have led to increased ecosystem carbon storage and might account for at least part of the "missing" sink required to balance the historical global carbon budget. To test this hypothesis, we employed a transient global terrestrial carbon cycle model of 0.5° spatial resolution. The model was driven with historical time series of monthly climate and the historical record of changes in atmospheric CO₂ concentration. The model's cumulative carbon sink for the period 19001988 accounts for about 68% of the estimated total "missing" sink. For the period 19501988, the model indicates that climate change and especially CO₂ fertilization may account for about 45% of the "missing" sink during that period.

Terrestrial ecosystems exchange a large amount of CO₂ with the atmosphere each year. Approximately 15% of the atmospheric C, or 120 Gt C, is assimilated in terrestrial-plant photosynthesis each year, and about an equal amount is returned to the atmosphere as CO₂ in plant respiration and the decomposition of plant litter and soil organic matter. The many processes that determine the magnitude of the carbon exchanges between the atmosphere and terrestrial ecosystems are sensitive to climate factors and changes in atmospheric CO₂. Time series of the

Fig. 1. Estimates net ecosystem production (NPP-R_H) changed by 2 Gt C globally
between 1982 and 1984 as a result of changing weather patterns produced by the
19821983 El Niño.

atmospheric CO₂ concentration since 1958 show both a long-term secular increase, largely due to fossil fuel burning, and shorter-term seasonal oscillations related to the activity of terrestrial biota in the northern hemisphere. In addition to these regular features, there are relatively large anomalies that have been attributed to transient imbalances in natural carbon exchange rates between the atmosphere and terrestrial ecosystems, perhaps caused by climate fluctuations such as the El Niño-Southern Oscillation (ENSO) phenomenon. It has been suggested that subtle and systematic changes in the net carbon fluxes as the result of climate variations or rising CO₂ concentrations over the past century may account for a substantial portion of the "missing" carbon sink of approximately 1 to 2 Gt C yr^-1 needed to balance the contemporary atmospheric CO₂ budget.

The rate at which carbon accumulates or is released from terrestrial ecosystems depends on not only the rate of physiological processes but also the size of various plant and soil pools, which have a wide range of turnover times. Changes in carbon storage in relatively slow turnover pools, especially wood and soil humus, determine whether terrestrial systems are net sources or sinks of carbon with respect to the atmosphere (Post et al. 1996). It is therefore necessary to keep track of not only the rates of carbon fixation, respiration and decomposition, and their fluctuation with environmental conditions, but also the size of the living and dead organic matter pools that are affected. Accordingly, we employ a global georeferenced compartment model of carbon in vegetation and soil (King et al. 1997; Post et al. 1997).

We partition the Earth's surface with a square lattice of 0.5 degree latitude by 0.5 degree longitude resolution. Each land cell is assigned to one of 31 ecosystem types according to the vegetation map of Matthews (1983) and to one of 106 soil types according to the FAO Soil Map of the World. For each vegetated cell carbon uptake as net primary production (NPP) is modeled with the Miami Model (Lieth 1975) modified to include a CO₂ response term based on theory and observation of leaf photosynthetic response that allows NPP to increase with increasing atmospheric CO₂. The calculated NPP is partitioned among four plant parts: leaf, branch, stem, and root. Litter production for each plant part is a function of life span parameters and the biomass of each plant part. The dynamics of litter and soil carbon are calculated with an implementation of the Rothamsted soil carbon turnover model (Jenkinson 1990).

We equilibrated and initialized the compartment model for each vegetated land cell using climate and atmospheric CO₂ concentrations appropriate for the end of the nineteenth century. We then simulated changes in NPP and carbon storage from 1900 to 1988 using monthly temperature and precipitation from observed weather records. The history of atmospheric CO₂ concentrations was prescribed using CO₂ measurements from the Siple Station ice core for years before 1959 and the atmospheric monitoring station at Mauna Loa Observatory thereafter.

Figure 2 shows modeled changes in terrestrial biospheric carbon of natural ecosystems, not significantly affected by land-use. There was a nearly linear increase in vegetation carbon

of about 40 Gt C (5%). Since the model's vegetation carbon pools are not directly affected by either temperature or precipitation, changes in vegetation carbon are the consequence of changes in NPP (Fig. 3a). Interannual variability in temperature and precipitation causes fluctuations in NPP on the order of 1.0 Gt C with occasional changes of 2.0 Gt C and even 3.0 Gt C (e.g., 19721975).

There is evidence of a decline in total soil organic carbon over the period 19101975. Soil carbon increases after 1975, resulting in a small net gain of 13.6 Gt C (1%) over the entire simulation. The loss of soil carbon is coincident with declining litter inputs (Fig. 3b), suggesting that the loss in soil C is a consequence of changes in litter inputs. The loss in soil carbon slows and then reverses itself as litter production increases. The rather abrupt increase in modeled soil organic carbon after 1975 (Fig. 2c) reflects accelerated CO₂ enhancement of NPP (Fig. 3a) and an abrupt increase in litter inputs to the soil system (Fig. 3b).

Climate-induced changes in decomposition rates may contribute to the decline in soil carbon, but increased decomposition rates during the period of soil carbon decline are not evident in changes in R_H, the carbon released as CO₂ during decomposition (Fig. 3c).

Changes in global total ecosystem carbon constitute net terrestrial biospheric exchange with the atmosphere. Our modeled estimate of this net biospheric flux, the difference between R_H and NPP, is shown in Fig. 4. Positive values indicate a biospheric source; negative values indicate a biospheric sink. The modeled net biospheric flux shows considerable interannual variability. Interannual differences in the modeled flux are commonly 0.5-1.0 Gt C, with occasional excursions of 2.0 Gt C (Fig. 1). The modeled biosphere frequently changes from source to sink or sink to source over periods of 1 to 3 years.

Figure 4 also shows an estimate of the net terrestrial biospheric flux derived by deconvolution of the historical atmospheric CO₂ record. The deconvolution generates the residual flux required to balance the observed atmospheric CO₂ record with industrial CO₂ emissions, net land-use emissions (Houghton and Hackler 1995), and modeled oceanic uptake estimated using a one-dimensional box-diffusion ocean model. Assuming that the residual flux is the result of terrestrial processes, the deconvolution provides an independent estimate of the historical net carbon exchange between the atmosphere and the terrestrial biosphere. In performing a deconvolution, some inputs are necessarily smoothed to prevent large fluctuations that arise from a lack of smoothness resulting from measurement uncertainties. There is no way to unsmooth the deconvolved estimate or smooth the results of the forward calculation of our model that would make them consistent. While this difference complicates our analysis the comparison between model and deconvolution is useful.

There is a general agreement in trends or temporal evolution of the natural biospheric flux (without land-use) over the 80-year period compared to the deconvolution. Allowing for interannual variability, the modeled flux with a land-use flux added agrees reasonably well with the deconvolution in the period of 1910 to 1930 and in the sink period of 19751980. In other periods modeled flux disagrees with the deconvolved estimate. The model predicts a relatively weak sink and perhaps a source during the period 19301950 when the deconvolution suggests an increasingly stronger sink. Moreover, the model predicts a sink that is too weak from 1950 to 1975 compared to the deconvolution.

Our simulations provide model evidence that the observed temporal pattern or evolution of net biospheric flux is partially attributable to a combination of climate and atmospheric CO₂ forcings and perhaps to climate variation alone. However, the model is unable to reproduce the temporal history and magnitude of the "missing" terrestrial sink derived by deconvolution. Assuming that the deconvolution is an accurate method of determining the historical net biosphere flux, then either the land-use term is too large for most of the period since 1930, or the modeled sink in response to changes in climate and atmospheric CO₂ is too small.

References

Houghton, R. A. and J. L. Hackler. 1995. Continental scale estimates of the biotic carbon flux from land cover change: 18501980, ORNL/CDIAC-79, Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, Oak Ridge, Tennessee.

Jenkinson, D. S. 1990. The turnover of organic carbon and nitrogen in soil. Phil. Trans. R. Soc. Lond. B 329:361368.

King, A. W., W. M. Post, and S. D. Wullschleger. 1997. The potential response of terrestrial carbon storage to changes in climate and atmospheric CO₂. Climatic Change 35:199227.

Lieth, H. 1975. Modelling the primary productivity of the world. pp. 237263. In H. Lieth and R. H. Whittaker (eds.), Primary Productivity of the Biosphere. Springer-Verlag, Berlin.

Matthews, E. 1983. Global vegetation and land use: New high resolution data bases for climate studies. J. Clim. Appl. Met. 22:474487.

Post, W. M., King, A. W., and S. D. Wullschleger 1996. Soil organic matter models and global estimates of soil organic carbon. pp. 201222. In P. Smith, J. Smith and D. Powlson (eds.), Evaluation of Soil Organic Matter Models Using Existing Long-Term Datasets. Springer-Verlag, Berlin.

Post, W. M., King, A. W., and S. D. Wullschleger. 1997. Historical variations in terrestrial carbon storage. Global Biogeochemical Cycles 11:99109.

Investigators

Wilfred M. Post
Oak Ridge National Laboratory
P.O. Box 2008
Bldg. 1000, MS-6335
Oak Ridge, Tennessee 37831-6335
(865) 576-3431
FAX: (865) 574-2232
E-mail: wmp@ornl.gov

Anthony W. King
Oak Ridge National Laboratory
P.O. Box 2008
Bldg. 1000, MS-6335
Oak Ridge, Tennessee 37831-6335
(865) 574-3436
FAX: (865) 574-2232
E-mail: awk@ornl.gov

Forrest Hoffman
Oak Ridge National Laboratory
P.O. Box 2008
Bldg. 1505, MS-6036
Oak Ridge, Tennessee 37831-6335
(865) 576-7680
E-mail: forrest@ornl.gov

Stan D. Wullschleger
Oak Ridge National Laboratory
P.O. Box 2008
Bldg. 1059, MS-6335
Oak Ridge, Tennessee 37831-6422
(865) 574-7389
E-mail: wullschlegsd@ornl.gov

Program Manager

Roger C. Dahlman
Office of Health and Environmental Research ER-74,
U.S. Department of Energy
19901 Germantown Road Germantown, Maryland, D.C. 20874
(301) 903-4951
FAX (301) 903-8519
Email: roger.dahlman@oer.doe.gov

Prepared by the Carbon Dioxide Information Analysis Center, OAK RIDGE NATIONAL LABORATORY, Oak Ridge, Tennessee 37831-6335 managed by LOCKHEED MARTIN ENERGY RESEARCH CORP. for the U.S. DEPARTMENT OF ENERGY under contract DE-AC05-96OR22464.

This electronic document was prepared by Marvel D. Burtis, CDIAC. (6/24/97)