Journal cover Journal topic
Geoscientific Model Development An interactive open-access journal of the European Geosciences Union
Geosci. Model Dev., 10, 2321-2332, 2017
https://doi.org/10.5194/gmd-10-2321-2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.
Methods for assessment of models
23 Jun 2017
A Bayesian posterior predictive framework for weighting ensemble regional climate models
Yanan Fan1, Roman Olson2, and Jason P. Evans3 1School of Mathematics and Statistics, UNSW, Sydney, Australia
2Department of Atmospheric Sciences, Yonsei University, Seoul, South Korea
3Climate Change Research Centre and ARC Centre of Excellence for Climate System Science, UNSW, Sydney, Australia
Abstract. We present a novel Bayesian statistical approach to computing model weights in climate change projection ensembles in order to create probabilistic projections. The weight of each climate model is obtained by weighting the current day observed data under the posterior distribution admitted under competing climate models. We use a linear model to describe the model output and observations. The approach accounts for uncertainty in model bias, trend and internal variability, including error in the observations used. Our framework is general, requires very little problem-specific input, and works well with default priors. We carry out cross-validation checks that confirm that the method produces the correct coverage.

Citation: Fan, Y., Olson, R., and Evans, J. P.: A Bayesian posterior predictive framework for weighting ensemble regional climate models, Geosci. Model Dev., 10, 2321-2332, https://doi.org/10.5194/gmd-10-2321-2017, 2017.
Publications Copernicus
Download
Short summary
We develop a novel and principled Bayesian statistical approach to computing model weights in climate change projection ensembles of regional climate models. The approach accounts for uncertainty in model bias, trend and internal variability. The weights are easily interpretable and the ensemble weighted models are shown to provide the correct coverage and improve upon existing methods in terms of providing narrower confidence intervals for climate change projections.
We develop a novel and principled Bayesian statistical approach to computing model weights in...
Share