Calibrating large-ensemble European climate projections using observational data

Abstract

Abstract. This study examines methods of calibrating projections of future regional climate using large single model ensembles (the CESM Large Ensemble and MPI Grand Ensemble), applied over Europe. The three calibration methods tested here are more commonly used for initialised forecasts from weeks up to seasonal timescales. The calibration techniques are applied to ensemble climate projections, fitting seasonal ensemble data to observations over a reference period (1920–2016). The calibration methods were tested and verified using an imperfect model approach using the historical/RCP 8.5 simulations from the CMIP5 archive. All the calibration methods exhibit a similar performance, generally improving the out-of-sample projections in comparison to the uncalibrated (bias-corrected) ensemble. The calibration methods give results that are largely indistinguishable from one another, so the simplest of these methods, namely Homogeneous Gaussian Regression, is used for the subsequent analysis. An extension to this method – applying it to dynamically decomposed data (in which the underlying data is separated into dynamical and residual components) – is also tested. The verification indicates that this calibration method produces more reliable and accurate projections than the uncalibrated ensemble for future climate over Europe. The calibrated projections for temperature demonstrate a particular improvement, whereas the projections for changes in precipitation generally remain fairly unreliable. When the two large ensembles are calibrated using observational data, the climate projections for Europe are far more consistent between the two ensembles, with both projecting a reduction in warming but a general increase in the uncertainty of the projected changes.

Combination of Decadal Predictions and Climate Projections in Time: Challenges and Potential Solutions

“…Our results suggest that the concatenation inconsistencies are primarily due to large differences in the predictions and projections ensemble spreads. To examine this further, we calibrated the MMEs using the variance inflation (VINF) method which scales the ensemble spread and signal using observed variability (see Doblas‐Reyes et al., 2005; O’Reilly et al., 2020, and Supporting Information for further information). VINF is applied to the climate projection MME from 1970 to 2014 and separately for each forecast year to the decadal prediction MME (using leave‐one‐out cross‐validation, Doblas‐Reyes et al.…”

Calibrating large-ensemble European climate projections using observational data

“…the coefficient d), so is sometimes referred to as "Nonhomogenous Gaussian Regression" (e.g. Wilks, 2006;Tippett and Barnston, 2008). EMOS represents a simplification over the VINF method because the the ensemble distribution is parameterised as Gaussian.…”