The November 2005 issue of GEWEX News (GEWEX is the Global Energy and Water Cycle Experiment) has an excellent article by T. Palmer of the ECMWF entitled âEnsemble Weather and Climate Predictionâ?.
His article states âthe representation of model uncertainty is a developing subjectâ?.
This is an important conclusion that needs to be widely recognized by the climate community.
He identifies three currently used methods to assess uncertainty including multi-model ensembles, perturbed-parameter ensembles, and stochastic physics. Multi-model ensembles have been utilized in multi-decadal retrospective predictions, such as in the CCSP report âTemperature Trends in the Lower Atmosphere: Understanding and Reconciling Differencesâ? . However, the use of perturbed-parameter ensembles (where uncertainties in the tunable parameters within model parameterizations are used to run the models with different values of these paramters), and stochastic physics (where the parameterizations include a statistical component) have not been completed with climate models. Of course, we also need to include all of the first order climate forcings and feedbacks, which, as we have discussed several times on this weblog, are not yet in the mult-decadal climate prediction models.
His article provides encouragement for model uncertainty assessments by the climate modeling community