54
Steven W. Running, Peter E. Thornton, Ramakrishna Nemani, and Joseph M. Glassy
daily CO2 flux, which is then coupled to an atmospheric mixing model to transport the land surface
fluxes into the atmosphere to match the flask sampling. This computation can be done for the few
dozen sampling sites that exist around the world,
providing some regional analysis of carbon source
and sink dynamics that can be very revealing of
biospheric dynamics (e.g., Keeling et al. 1996).
Nemery et al. (1999) found that terrestrial biospheric models quantify northern hemisphere CO 2
seasonality adequately, but all models miss the
southern hemisphere cycle, possibly because of
poor representation of tropical rain-green phenology. Of course, atmospheric CO 2 is not vegetation
production, so again this technique cannot directly
validate NPP.
A new technique for large-scale carbon cycle
validation incorporates eddy covariance flux towers
that detect terrestrial CO 2 fluxes continuously (Baldocchi et al. 1996; Goulden et al. 1996). A coordinated network of 80 towers currently operates
globally called FLUXNET, and CO 2 fluxes are
measured year around. Again, these fluxes must be
transformed into vegetation NPP to be a direct validation. The carbon allocation theory needed to
transform a tower net ecosystem exchange (NEE)
,,-..... 3000
....
D
ID
>.
2000
"' "
E
E
..., 1000
c
.91
()
.~
;.::
0
"Q)
0
()
Q) - 1000
()
c
D
0
.0
- 2000
1000 - 1500
LID
....
0
< 0
~ -3000
into aboveground NPP equivalent to the MODIS
variable requires partitioning GPP into respiration
and growth components, and separating soil heterotrophic respiration (Waring and Running 1998).
Additionally, the towers only sample roughly 1 to
5 km 2 of land area, so spatial scaling procedures
must be established to even approach a global analysis of NPP (Running et al. 1999). Nevertheless,
these NEE data provide the highest precision and
most temporally continuous carbon cycle measurements available (Kimball et al. 1997b).
The only way to generate complete coverage of
NPP in both space and time is by model computations that incorporate some of these measurements. These global NPP models are initialized and
operated with a variety of real measurements of the
terrestrial biophysical system, but certainly cannot
be called a direct validation. Some use regular satellite mapping of global LAI, most use global
weather station data, and some use flux and/or isotopic data to define the ecosystem. These modeled
NPP allow analyses that usually are not possible
with individual field data. For example, the environmental space occupied by global NPP gives an
opportunity to evaluate the general climatic controls on productivity (Fig. 3.6) (Churkina and RunIn Climate Space
-1 0
0
10
20
annual mean temperature (deg C)
30
FIGURE 3.6. An analysis of global annual NPP in climate space, defined by annual temperature and water balance,
to understand the range and limitation of NPP by climatic constraints. (From Churkina and Running [1998].)
Steven W. Running, Peter E. Thornton, Ramakrishna Nemani, and Joseph M. Glassy
daily CO2 flux, which is then coupled to an atmospheric mixing model to transport the land surface
fluxes into the atmosphere to match the flask sampling. This computation can be done for the few
dozen sampling sites that exist around the world,
providing some regional analysis of carbon source
and sink dynamics that can be very revealing of
biospheric dynamics (e.g., Keeling et al. 1996).
Nemery et al. (1999) found that terrestrial biospheric models quantify northern hemisphere CO 2
seasonality adequately, but all models miss the
southern hemisphere cycle, possibly because of
poor representation of tropical rain-green phenology. Of course, atmospheric CO 2 is not vegetation
production, so again this technique cannot directly
validate NPP.
A new technique for large-scale carbon cycle
validation incorporates eddy covariance flux towers
that detect terrestrial CO 2 fluxes continuously (Baldocchi et al. 1996; Goulden et al. 1996). A coordinated network of 80 towers currently operates
globally called FLUXNET, and CO 2 fluxes are
measured year around. Again, these fluxes must be
transformed into vegetation NPP to be a direct validation. The carbon allocation theory needed to
transform a tower net ecosystem exchange (NEE)
,,-..... 3000
....
D
ID
>.
2000
"' "
E
E
..., 1000
c
.91
()
.~
;.::
0
"Q)
0
()
Q) - 1000
()
c
D
0
.0
- 2000
1000 - 1500
LID
....
0
< 0
~ -3000
into aboveground NPP equivalent to the MODIS
variable requires partitioning GPP into respiration
and growth components, and separating soil heterotrophic respiration (Waring and Running 1998).
Additionally, the towers only sample roughly 1 to
5 km 2 of land area, so spatial scaling procedures
must be established to even approach a global analysis of NPP (Running et al. 1999). Nevertheless,
these NEE data provide the highest precision and
most temporally continuous carbon cycle measurements available (Kimball et al. 1997b).
The only way to generate complete coverage of
NPP in both space and time is by model computations that incorporate some of these measurements. These global NPP models are initialized and
operated with a variety of real measurements of the
terrestrial biophysical system, but certainly cannot
be called a direct validation. Some use regular satellite mapping of global LAI, most use global
weather station data, and some use flux and/or isotopic data to define the ecosystem. These modeled
NPP allow analyses that usually are not possible
with individual field data. For example, the environmental space occupied by global NPP gives an
opportunity to evaluate the general climatic controls on productivity (Fig. 3.6) (Churkina and RunIn Climate Space
-1 0
0
10
20
annual mean temperature (deg C)
30
FIGURE 3.6. An analysis of global annual NPP in climate space, defined by annual temperature and water balance,
to understand the range and limitation of NPP by climatic constraints. (From Churkina and Running [1998].)