Connecting Science

1
Java Climate Model,
Connecting Science Policy, from Emissions to
Impacts
Try it yourself, on the Web climatechange.unep
.net/jcm/ www.chooseclimate.org/jcm/
www.climate.unibe.ch/matthews/jcm/
Dr Ben Matthews ben_at_chooselimate.org Currently
at Klima UmweltPhysik, Univ Bern DEA-CCAT
/Energimiljoradet Copenhagen, UNEP-GRID Arendal
Norway, UCL Louvain-la-Neuve Belgium
The Java Climate Model enables anybody to explore
many climate policy options and scientific
uncertainties, simply by adjusting parameter
controls with a mouse in a web browser, and
observing the change on linked plots. The instant
mechanical response demonstrates cause and
effect and the sensitivity to assumptions, risk
and value judgements.
JCM also developed with
Everything is interconnected! The cause-effect
links may be demonstrated by a flowchart which
adjusts depending on the question you ask. In
the example below, the red arrows show the effect
of dragging the temperature stabilisation control
to 2C above preindustrial (1765). This goal is
achieved by an iterative feedback from climate to
mitigation policy, in this case reducing the
emissions of all greenhouse gases (except CFCs)
and aerosols by the same fraction relative to
SRES A1B. This requires CO2 stabilisation at
approximately 500ppm.
Core global science (left / base of flowchart ,
all plots from 1750 to 2300). The model
includes an efficient java implementation of
simple carbon and climate models, as used for
IPCC-TAR predictions. The carbon cycle,
atmospheric chemistry, and radiative forcing are
based on the Bern CC-model (the terrestrial
biosphere sink is now being updated). The
temperature and sea-level rise calculations are
based on the Wigley-Raper model with parameters
fitted to GCM data (in this case HadCM2), see
IPCCTAR WG1appx 9.1. There is a feedback from
climate to carbon, the models are fully coupled
within one time-loop. The slow penetration of
carbon and heat into the oceans is calculated by
an eigenvector method from DEA-CCAT. The
correspondence with published IPCC data can be
illustrated by circles superimposed on the plot
(see the example below, which shows SRES B1).
Regional plots (top / right of flowchart) These
show data for 12 regions (colours as map) from
1900-2100. SRES scenarios define socio-economic
data (from IMAGE). These may be combined with
various distribution formulae to calculate
regional CO2 emissions quotas, and also
abatement and accumulated atmospheric CO2
(emissions minus sinks) from each region In
this example, CO2 emissions follow Kyoto targets
to 2012, then converge to equal per-capita levels
by 2040, while the population, GDP, and abatement
baseline are fixed by SRES A1B. The regional
climate map (left) shows temperature change from
HadCM2 GCM data, scaled to global average.
Eventually, to complete the circle and make the
models more relevant, we need to develop feedback
from specific local climate impacts, to local
mitigation / adaptation policy. However, regional
climate models now take months to run, so
achieving this interactively will be a challenge!
Interface, documentation, code You can adjust the
plot layout, scales, and complexity level.
Information about controls, curves, and units
pops up as the mouse moves about. Extensive web
documentation is dynamically linked to the model,
with automatic demo-wizards illustrating key
questions. Alternatively one click in a plot or
flowchart can call documentation or source code
for a specific control or module. The model
works on most web browsers and computers, is very
compact (lt200k) downloading in seconds, and also
works offline. It has an adaptable yet efficient
structure of interacting modules, which calculate
only when both needed by plots and changed by
controls. It may also be instructed by
text-based scripts, for automatic demonstrations
or batch calculations.
A Quantitative Framework for Global
Dialogue Steering our ship to avoid dangerous
anthropogenic interference in the climate system
(UNFCCC article 2) requires balancing many risk
and value judgements, and computer models may
help provide a quantitative framework to help
resolve the complex interactions. Yet these
remain a mysterious black box to all but a few
experts, whilst to effectively implement any
global agreement requires the engagement of many
citizens around the world. So we need some
democratisation of climate science. The
ultimate integrated assessment model will
remain the global network of human heads, and the
Java Climate Model is designed to assist this
dialogue. The code is also internationalised so
the pop-up info and labels can be translated into
any language (six are already available), and the
model is compact and fast, considering users with
slower computers and connections. Several users
may share the same model by remote control
across the internet. This works efficiently,
sending only parameters as they change. However
synchronisation may be a challenge. Alternatively
users may record a script of model events to
illustrate a point, and post this on the web for
viewing by others later.