308
Air Pollution and Turbulence: Modeling and Applications
Inversion techniques: Further development of inversion techniques can be
expected, both to refi ne the type of calculations reviewed here and to accommodate
the new types of data. A few new techniques are noted in Section 11.2.4. Some other
likely future directions include the increased use of nonlinear estimation (Enting,
2002, Chapter 12). Linear estimation is appropriate when the model relations are
linear and the statistics are multivariate Gaussian, as in the classic “synthesis”
approach to estimating fl uxes. As more complex systems are analyzed, these simplifying conditions are likely to prove less and less adequate as working approximations. Applications involving varying forms of data assimilation are also likely to
form an increasing part of carbon cycle research (Wang et al., 2009). In these various developments, the use of adjoint modeling (Giering, 2000) is likely to become
increasingly important.
Evolving objectives: As noted in Section 11.4.1, much of the current interest in the
carbon cycle involves the coupling between the carbon cycle and the physical climate
system. Such coupling implies feedbacks that could exacerbate the ongoing anthropogenic warming. A range of initial studies are being undertaken in the context of
the C4MIP intercomparison (Fung et al., 2000; Rayner, 2001; Friedlingstein et al.,
2006). Carbon cycle inversions may have a future role in detecting the “fi ngerprints”
of such feedbacks when and if they occur. In the more immediate future, carbon
cycle inversions are likely to focus on the calibration of the models that are being
used to project climate change in the presence of such feedbacks. As discussed in
Section 11.2.4, such “process inversion” may often involve “data assimilation” using
proxy data to estimate, via model relations, the forcing of the carbon system. It is
also possible that the links between photosynthesis and plant water use may mean
that carbon data provide information about terrestrial water balance, leading to possible applications in synoptic and seasonal forecasting. In these various inversion
problems, the multiple interacting scales, a characteristic of complex systems, (c.f.,
Falkowski et al., 2000), lead to a new class of mathematically challenging inverse
problems. As a further step, the Global Carbon Project plan (Global Carbon Project,
2003) envisages modeling of the coupled interactions of the climate system with
human systems. This will require a major realignment of modeling approaches.
11.6 CONCLUDING REMARKS
In the concluding chapter of my book, I noted the rapid growth in the fi eld of trace
gas inversions. This rapid growth has continued and the pace is accelerating. New
data streams will stimulate further growth, particularly once satellite data become
available.
An analysis of the science of ozone depletion and the ozone hole (Christie, 2000)
notes that for both the British Halley Bay observations and the NASA satellite retrievals, limitations in data processing and analysis led to a delay in recognition of the
ozone hole for a number of years after it appeared. It was suggested that this refl ected
the less-developed state of computer systems in the early 1980s and that a similar
situation is now less likely to occur. Compared to the 1980s, raw computing power
is indeed less of a problem now, although the projected volume of raw satellite data
is formidable. There would seem to be a high risk of a gap between observational
© 2010 by Taylor and Francis Group, LLC
Air Pollution and Turbulence: Modeling and Applications
Inversion techniques: Further development of inversion techniques can be
expected, both to refi ne the type of calculations reviewed here and to accommodate
the new types of data. A few new techniques are noted in Section 11.2.4. Some other
likely future directions include the increased use of nonlinear estimation (Enting,
2002, Chapter 12). Linear estimation is appropriate when the model relations are
linear and the statistics are multivariate Gaussian, as in the classic “synthesis”
approach to estimating fl uxes. As more complex systems are analyzed, these simplifying conditions are likely to prove less and less adequate as working approximations. Applications involving varying forms of data assimilation are also likely to
form an increasing part of carbon cycle research (Wang et al., 2009). In these various developments, the use of adjoint modeling (Giering, 2000) is likely to become
increasingly important.
Evolving objectives: As noted in Section 11.4.1, much of the current interest in the
carbon cycle involves the coupling between the carbon cycle and the physical climate
system. Such coupling implies feedbacks that could exacerbate the ongoing anthropogenic warming. A range of initial studies are being undertaken in the context of
the C4MIP intercomparison (Fung et al., 2000; Rayner, 2001; Friedlingstein et al.,
2006). Carbon cycle inversions may have a future role in detecting the “fi ngerprints”
of such feedbacks when and if they occur. In the more immediate future, carbon
cycle inversions are likely to focus on the calibration of the models that are being
used to project climate change in the presence of such feedbacks. As discussed in
Section 11.2.4, such “process inversion” may often involve “data assimilation” using
proxy data to estimate, via model relations, the forcing of the carbon system. It is
also possible that the links between photosynthesis and plant water use may mean
that carbon data provide information about terrestrial water balance, leading to possible applications in synoptic and seasonal forecasting. In these various inversion
problems, the multiple interacting scales, a characteristic of complex systems, (c.f.,
Falkowski et al., 2000), lead to a new class of mathematically challenging inverse
problems. As a further step, the Global Carbon Project plan (Global Carbon Project,
2003) envisages modeling of the coupled interactions of the climate system with
human systems. This will require a major realignment of modeling approaches.
11.6 CONCLUDING REMARKS
In the concluding chapter of my book, I noted the rapid growth in the fi eld of trace
gas inversions. This rapid growth has continued and the pace is accelerating. New
data streams will stimulate further growth, particularly once satellite data become
available.
An analysis of the science of ozone depletion and the ozone hole (Christie, 2000)
notes that for both the British Halley Bay observations and the NASA satellite retrievals, limitations in data processing and analysis led to a delay in recognition of the
ozone hole for a number of years after it appeared. It was suggested that this refl ected
the less-developed state of computer systems in the early 1980s and that a similar
situation is now less likely to occur. Compared to the 1980s, raw computing power
is indeed less of a problem now, although the projected volume of raw satellite data
is formidable. There would seem to be a high risk of a gap between observational
© 2010 by Taylor and Francis Group, LLC