Inversion of Atmospheric CO 2 Concentrations
305
statistical fi t spans a somewhat wider range than the cases considered by
Tans et al. (1990).
3. Skin correction This repeats case 2, with a correction for the skin effect
identifi ed by Robertson and Watson (1992) to the priors for northern and
tropical oceans.
4. Inverse The fi nal line represents an inversion with very loose priors, and
using neither the total ocean constraint nor the pCO 2 data. As indicated by
the following two cases, additional constraints can greatly infl uence the
fi t. A wide range of ocean total uptake is consistent with the data. (The
original Bayesian synthesis calculations (Enting et al., 1995) used 13 C data
to constrain the land–ocean partitioning.)
11.4.3 RESULTS FROM INVERSIONS
Heimann et al. (2004) reviewed CO 2 inversions and noted that much of the smallscale information is really coming from the priors. Conversely, on the very largest
scales (i.e., global totals) inversions contribute relatively little to land–ocean partitioning (see Enting, 2002, Fig. 13.1). On the continental scale, inversions can produce
useful information. A summary of results is given below, including results from the
TransCom control experiment (Gurney et al., 2003), which was a 22-region synthesis
inversion of monthly mean data covering 1992–1996.
Ocean sink Section 11.4.2 describes discrepancies in estimates of ocean carbon
uptake produced circa 1990, and the resolution through consistent budgeting.
There were also, slightly later, two different analyses based on 13 C that also differed between high ocean uptake (Quay et al., 1992) and low ocean uptake (Tans et al.,
1993). A consistent 13 C-based budget was presented by Heimann and Maier-Reimer
(1996). For an overview of isotopic budgeting calculations, see Enting (2002, Section
10.4), which identifi es the Quay et al. budget as a “storage” budget and the Tans et al.
budget as a fl ux budget. The TransCom result is a sink of −1.5 ± 1.1 GtC y −1 .
The overall result is that inversions have played only a limited role in determining
the net ocean uptake. In spite of the concept that interhemispheric differences are
strongly correlated with land–ocean differences, the spatial gradient is only a poor
proxy for process differences. However, as discussed below, inversion studies have
put useful constraints on southern hemisphere ocean fl uxes. To a large extent, some
of the strongest constraints on estimates of ocean carbon uptake come from measurements of O 2 :N 2 ratios. What these determine, when combined with CO 2 trends,
is the changes of oxidized carbon (essentially storage in the ocean) versus reduced
carbon (primarily in terrestrial biota). As such it is a “storage” budget rather than a
fl ux budget.
Tropical biota: The appreciation of the large carbon releases from land-use change
in tropical regions brought the carbon budget discrepancies (the issue of the so-called
missing sink) into particular prominence. The ability to estimate net tropical fl uxes
from global inversions is severely hampered by (a) the shortage of sampling sites in
the relevant regions and (b) the nature of the atmospheric circulation where tropical air is transported vertically, so that any anomalous concentration signal makes
little contribution to concentration gradients at tropical latitudes. Early mass-balance
© 2010 by Taylor and Francis Group, LLC
305
statistical fi t spans a somewhat wider range than the cases considered by
Tans et al. (1990).
3. Skin correction This repeats case 2, with a correction for the skin effect
identifi ed by Robertson and Watson (1992) to the priors for northern and
tropical oceans.
4. Inverse The fi nal line represents an inversion with very loose priors, and
using neither the total ocean constraint nor the pCO 2 data. As indicated by
the following two cases, additional constraints can greatly infl uence the
fi t. A wide range of ocean total uptake is consistent with the data. (The
original Bayesian synthesis calculations (Enting et al., 1995) used 13 C data
to constrain the land–ocean partitioning.)
11.4.3 RESULTS FROM INVERSIONS
Heimann et al. (2004) reviewed CO 2 inversions and noted that much of the smallscale information is really coming from the priors. Conversely, on the very largest
scales (i.e., global totals) inversions contribute relatively little to land–ocean partitioning (see Enting, 2002, Fig. 13.1). On the continental scale, inversions can produce
useful information. A summary of results is given below, including results from the
TransCom control experiment (Gurney et al., 2003), which was a 22-region synthesis
inversion of monthly mean data covering 1992–1996.
Ocean sink Section 11.4.2 describes discrepancies in estimates of ocean carbon
uptake produced circa 1990, and the resolution through consistent budgeting.
There were also, slightly later, two different analyses based on 13 C that also differed between high ocean uptake (Quay et al., 1992) and low ocean uptake (Tans et al.,
1993). A consistent 13 C-based budget was presented by Heimann and Maier-Reimer
(1996). For an overview of isotopic budgeting calculations, see Enting (2002, Section
10.4), which identifi es the Quay et al. budget as a “storage” budget and the Tans et al.
budget as a fl ux budget. The TransCom result is a sink of −1.5 ± 1.1 GtC y −1 .
The overall result is that inversions have played only a limited role in determining
the net ocean uptake. In spite of the concept that interhemispheric differences are
strongly correlated with land–ocean differences, the spatial gradient is only a poor
proxy for process differences. However, as discussed below, inversion studies have
put useful constraints on southern hemisphere ocean fl uxes. To a large extent, some
of the strongest constraints on estimates of ocean carbon uptake come from measurements of O 2 :N 2 ratios. What these determine, when combined with CO 2 trends,
is the changes of oxidized carbon (essentially storage in the ocean) versus reduced
carbon (primarily in terrestrial biota). As such it is a “storage” budget rather than a
fl ux budget.
Tropical biota: The appreciation of the large carbon releases from land-use change
in tropical regions brought the carbon budget discrepancies (the issue of the so-called
missing sink) into particular prominence. The ability to estimate net tropical fl uxes
from global inversions is severely hampered by (a) the shortage of sampling sites in
the relevant regions and (b) the nature of the atmospheric circulation where tropical air is transported vertically, so that any anomalous concentration signal makes
little contribution to concentration gradients at tropical latitudes. Early mass-balance
© 2010 by Taylor and Francis Group, LLC