final_program_abstracts[1]

11 IMSC Session Program
How good do palaeoclimate reconstructions of last 1000
years need to be to usefully constrain climate model
parameters?
Tuesday - Plenary Session 3
Timothy J. Osborn, Douglas Maraun, Sarah C.B. Raper and Keith R. Briffa
Recent work has used reconstructions of past temperature variability to estimate or
constrain climate model parameters such as climate sensitivity (e.g., Hegerl et al.,
2006) or carbon cycle feedback (Frank et al., 2009). Our study takes a step back and
uses synthetic data, constructed using output from climate model simulations, to
explore how the potential constraints on climate model parameters depend on the
uncertainty associated with temperature reconstructions. We also consider the
influences of internally-generated climate variability, which tends to mask the
response to external climate forcings, and of the uncertainty associated with past
climate forcings. The results are likely to be optimistic compared with our ability to
constrain equivalent parameters of the real climate system, because the experimental
design makes a number of assumptions about prior knowledge of the synthetic
climate, though the influence of these assumptions can be quantified.
Synthetic temperature reconstructions were generated by combining a simulation of
externally-forced climate change (produced using an energy balance climate model)
with several hundred realizations of internal climate variability (diagnosed from a
multi-century control simulation with a GCM-based climate model). An iterative
optimisation algorithm, involving thousands of simulations with the energy balance
climate model, was adopted to obtain a maximum-likelihood estimate of unknown
model parameters (climate sensitivity and parameters that determine the timescaledependence
of ocean effective heat capacity). The approach was also extended to
estimate the amplitudes of past forcings to explore the relative importance of accurate
temperature reconstructions and uncertainty in estimates of climate forcings.
These experiments reveal a range of results: (i) climate reconstructions for the preinstrumental
period have the potential to provide additional constraints upon climate
model parameters (and hence future climate predictions) but this potential is limited
by the need for accurate estimates of past forcing factors; (ii) the response to past
volcanic eruptions appears to provide some constraint upon the lower values of
climate sensitivity, but less so for the higher values; (iii) observing the response to
forcings that operate across multiple time scales (e.g., volcanic and solar variations
together) is necessary if both the climate sensitivity and ocean heat uptake need to be
estimated, especially if the strengths of the forcings are also uncertain.
Abstracts 88