The Impact of Bias Correction and Model Selection on Passing Temperature Thresholds

Knowledge of when specific global or local temperature levels are reached is important for decision makers in that it provides a time frame over which adaptation strategies for temperature-related climate impacts need to be put in place. The time frame varies depending on the adaptation strategy but can range from a few years to the order of decades. Climate models, however, show a high degree of uncertainty in the timing of passing specific warming levels, limiting their use in adaptation policy development. This study examines the impact of two approaches, which may reduce the uncertainty in modeled timing of reaching specific warming levels. First, the use of different performance metrics to preferentially weight model ensembles and second, the application of four bias correction approaches. Using the Coupled Model Intercomparison Project phase 5 simulations of the Representative Concentration Pathways, our results show that both selecting models based on performance and bias correcting model data reduce the spread in timing of specific warming levels reached in the first half of the century by up to 50% in some regions. This implies the potential of these approaches to support adaptation planning.