Geoscience Reference
In-Depth Information
Fourth, IA can play an important role in helping to translate and communicate
uncertainty by showing how differences in worldviews infl uence the framing of sci-
entifi c problems and choices about how social systems respond to climate change.
In this way, and linked to the third point above, IA can help identify and prioritise
critical gaps in scientifi c knowledge.
A very large number of Integrated Assessments have been undertaken and a large
number of models have been created and improved over many years of work. Over
time, key submodels are added, refi ned and their links to other models are improved.
Some of the classic examples include:
• Mackenzie Basin Impact Study, which examined the interactions between
climate change, regional development and management responses in Canada's
Mackenzie Basin.
• POLESTAR was developed in the US, specifi cally to explore the interactions
between natural and socio-political systems. Driving forces that underpinned the
various scenarios included population growth, development strategies and soci-
etal responses.
•
GCAM, an IAM which represented the interactions between atmospheric com-
position, human active, climate and sea level and ecosystems.
•
IMAGE 2.0 focuses on the interactions between socio-economic processes, land
cover, atmospheric and climate process, ecological and economic impacts and
interventions in a series of broader 'Pressure-state-impact-response-loops'.
Looking internationally, Rotmans and Dowlatabadi identifi ed a major divide in
the underlying approaches between European approaches, more focused biosphere-
climate oriented models, and North American ones more focused on macro-
economic models. Of course, there are exceptions, but the broad difference does
refl ect interesting national differences in the framing of the climate problem as a
whole.
In summarising some of the limitations and drawbacks of IAM, Rotmans and
Dowlatabadi point to a number of problems. Overly complex model structures
result in researchers learning more about the model than reality. Many IAMs have
unacceptably high levels of aggregation and do not include random behaviour. In
addition, there is often inadequate treatment of uncertainty within climate models,
limited verifi cation and validation and limitations on the methods used within the
model. All these factors reduce the credibility of models within the wider scientifi c
community. Moreover, despite all the caveats attached to models, they are still often
treated as predictive truth machines, rather than heuristic tools or abstractions from
much more complex systems.
Many of these limitations and drawbacks refl ect the underlying challenge of
climate change: it demands that researchers seek to represent systems where they
know that they do not understand many of the dynamics of the system. Judged by
the standard canons of science, this is the wild west of research: frontier work with
crude tools cobbled together into an odd looking toolkit. The standard response
from IA modellers is that by the time we are able to fully specify the problem, it
will be too late to take any meaningful action.
Nonetheless IA practitioners have given serious consideration to issues of calibra-
tion, validation, and uncertainty. The challenge in all three domains is that the
systems are open and that there is no baseline or benchmark against which to
