Challenges in Combining Projections from Multiple Climate Models

Abstract

Recent coordinated efforts, in which numerous general circulation climate models have been run for a common set of experiments, have produced large datasets of projections of future climate for various scenarios. Those multimodel ensembles sample initial conditions, parameters, and structural uncertainties in the model design, and they have prompted a variety of approaches to quantifying uncertainty in future climate change. International climate change assessments also rely heavily on these models. These assessments often provide equal-weighted averages as best-guess results, assuming that individual model biases will at least partly cancel and that a model average prediction is more likely to be correct than a prediction from a single model based on the result that a multimodel average of present-day climate generally outperforms any individual model. This study outlines the motivation for using multimodel ensembles and discusses various challenges in interpreting them. Among these challenges are that the number of models in these ensembles is usually small, their distribution in the model or parameter space is unclear, and that extreme behavior is often not sampled. Model skill in simulating present-day climate conditions is shown to relate only weakly to the magnitude of predicted change. It is thus unclear by how much the confidence in future projections should increase based on improvements in simulating present-day conditions, a reduction of intermodel spread, or a larger number of models. Averaging model output may further lead to a loss of signal—for example, for precipitation change where the predicted changes are spatially heterogeneous, such that the true expected change is very likely to be larger than suggested by a model average. Last, there is little agreement on metrics to separate “good” and “bad” models, and there is concern that model development, evaluation, and posterior weighting or ranking are all using the same datasets. While the multimodel average appears to still be useful in some situations, these results show that more quantitative methods to evaluate model performance are critical to maximize the value of climate change projections from global models.

A review on regional dynamical downscaling in intraseasonal to seasonal simulation/prediction and major factors that affect downscaling ability

“…However, based on the discussions in Sections 6.1 and 2.1, the RCM is very likely to mimic the imposed GCM LBC's trend and temporal variability, and the deficiency in GCM future prediction will be transferred to the RCM. Knutti et al (2010) have extensively discussed the challenges in combining projections from multiple GCM future predictions and concluded that extracting policy-relevant information is difficult. Among these challenges are that model skill in simulating present-day climate conditions is shown to relate only weakly to the magnitude of predicted change, and quantifying uncertainties from ensembles of climate models is difficult.…”

“…Boberg and Christensen (2012), in a central Mediterranean study, demonstrate that projections of intense mean summer warming partly result from model deficiencies, and after correcting the biases, the Mediterranean summer temperature projections in an ensemble mean are reduced by up to one degree, on average by 10-20%. In another approach, McSweeney et al (2012) demonstrated the importance of employing a well-considered sampling strategy to select GCMs used for LBCs based on Knutti et al (2010) and others' recommendations. They first examine whether any GCMs should be eliminated for LBCs because of significant deficiencies in their simulation of current climate for that region.…”

Uncertainties in the timing of unprecedented climates

“…While the estimate of the ensemble-mean becomes more precise with larger ensemble size, natural fluctuations of the climate (such as El Niño) dictate that the future evolution of climate will not behave like the mean, but as a single realization from a range of outcomes 5,7 . The use of σ/√N greatly underestimates 8 this irreducible uncertainty, as well as the climate-response uncertainty given by the inter-model spread, and is therefore inappropriate for use in emergence estimates.…”

Smoke consequences of new wildfire regimes driven by climate change

“…With the computational burdens of GCMs, combinatorial explosion is a real danger, so bounds must always be set on the number of combinations used. In general, quantitative evaluation of ensemble methods is still at an early stage, with limitations including the use of equal-weighted averages [but see Mote and Salathé, 2010], the necessarily small numbers of models used, the absence of extreme behavior emerging from averages, and the lack of agreement on what is a good metric for evaluation [Knutti et al, 2010].…”