Models of physical processes in the ocean and atmosphere provide much of our
current understanding of future climate change. They incorporate the contributions
of atmospheric dynamics and thermodynamics through the methods of computational
fluid dynamics. This approach was initially developed in the 1950s to provide
an objective numerical approach to weather prediction. It is sometimes forgotten
that the early development of "supercomputers" at that time was motivated
in large part by the need to solve this problem. In the 1960s, versions of these
weather prediction models were developed to study the "general circulation"
of the atmosphere, i.e., the physical statistics of weather systems satisfying
requirements of conservation of mass, momentum, and energy. To obtain realistic
simulations, it was found necessary to include additional energy sources and
sinks: in particular, energy exchanges with the surface and moist atmospheric
processes with the attendant latent heat release and radiative heat inputs.
Development of models for the general circulation of the ocean started later,
but has proceeded in a similar manner. Models that deal with the physics of
the oceans have been developed and linked to models of the atmospheric system.
Within ocean models, the inclusion of geochemical and biological interactions
has begun, with a focus upon the carbon cycle. Since the late 1960s, the geochemical
aspects of the carbon cycle have been included in low-dimensional box models.
More recently, including the carbon chemistry system in general circulation
models has simply been a question of allocation of computing resources. Modelling
of the biological system, however, has been more challenging, and it has only
been recently that primitive ecosystem models have been incorporated in global
general circulation ocean models. Even though progress has been significant,
much remains to be done. Eddy-resolving ocean models with chemistry and biology
need to be tested and validated in a transient mode, and the prognostic aspects
of marine ecosystems including nutrient dynamics need greater attention at basin
and global scales.
Model development for the ocean and atmosphere has had a fundamental theoretical
advantage: it is based on firmly established hydrodynamic equations. At present
there is less theoretical basis for a "first principles" development
of the dynamical behaviour of the terrestrial system. There is a need to develop
a fundamental methodology to describe this very heterogeneous and complex system.
For the moment, it is necessary to rely heavily upon parametrizations and empirical
relationships. Such reliance is data intensive and hence independent validation
of terrestrial system models is problematical. In spite of these difficulties,
a co-ordinated strategy has been developed to improve estimates of terrestrial
primary productivity and respiration by means of measurement and modelling.
The strategy has begun to yield dividends. Techniques from statistical mechanics
have been wedded to biogeochemistry and population ecology, yielding new vegetation
dynamic models. Global terrestrial models at meso-spatial scales (roughly 50
km grids) now exist which capture complex ecophysiological processes and ecosystem
Expanded efforts are needed in these domain-specific models. In the ocean,
we need to consider better the controls on thermohaline circulation, on potential
changes in biological productivity, and on the overall stability of the ocean
circulation system. Within terrestrial systems the question of the carbon sink-source
pattern is central: what is it and how might it change? Connected to this question
is the continued development of dynamic vegetation models, which treat competitive
processes within terrestrial ecosystems and their response to multiple stresses.
And for the atmosphere, a central question has been, is, and likely will be
the role of clouds. Also, there is a corresponding non-linearity associated
with change in the distribution and extent of sea ice. Further increased efforts
will be needed in linking terrestrial ecosystems with the atmosphere, the land
with the ocean, the ocean (and its ecosystems) with the atmosphere, the chemistry
of the atmosphere with the physics of the atmosphere, and finally linking the
human system to them all. Such models will also need to be able to highlight
different regions with increased spatial and temporal detail.
Models, however, depend upon high quality data. Data allow hypotheses about processes and their linkages to be rejected or to be given increased consideration. Giving formal (e.g., quantitative) expression to processes is at the heart of the scientific enterprise. Such expressions reflect our knowledge and form the basis for models. Models are simply formal expressions of processes and how they fit together. And all rest upon data. Models are of limited use without observations; the value of observations increases by interaction with models. Systematic global observations are an essential underpinning of research to improve understanding of the climate system. For numerous applications in climate-impact research, information about the complex nature of the system is needed. Unfortunately, there continue to be justifiable concerns about the loss of some monitoring of climate parameters and deterioration of coverage. There is a basic need for more observations with better coverage, higher accuracy, and with increased availability. This overriding importance of data has been recognised repeatedly in the past and in this volume (e.g., Chapter 2, Section 2.8; Chapter 3, Section 3.5; Chapter 4, Section 4.2; Chapter 6, Section 6.14; Chapter 11, Section 11.6.1 and Chapter 12, Section 12.4), and there are reasons for guarded optimism on the issue of data even though there are also significant reasons for concern. One such reason for tempered optimism is the plan for and beginning implementation of global observing systems such as the Global Climate Observing System (GCOS), Global Ocean Observing System (GOOS), and Global Terrestrial Observing System (GTOS). However plans in themselves do not produce data, and data that are not accessible are of limited value. The issue of data remains central for progress.
Other reports in this collection