We are using the citizen science regional climate modelling project weather@home to perform large ensembles of the different experiments described below. weather@home is a project using distributed computing to run very large ensembles of simulations with a regional model, the Met Office’s HadRM3P, nested in the global atmosphere-only model HadAM3P. Massey et al. (submitted) described the model setup and validated the regional model output. Figure 13 in Massey et al. (submitted) shows that HadRM3P produces realistic precipitation amounts for December-January-February (abbreviated as DJF from here on), the season of interest for this study.
We will be running two different experiments: one to represent the current winter as it was observed, and one to represent the current winter in the world that might have been without human-induced climate change. For the “current winter as observed” experiment, the atmospheric model uses observed sea surface temperature data from December 2013 and January and February 2014 from the Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) dataset (Stark et al. 2007; Donlon et al. 2012) and present day atmospheric gas concentrations to simulate weather events that are possible given the observed climate conditions. Figure 1 shows the differences between the 2013/2014 winter SSTs compared to the winter climatology (average over all years) of the OSTIA dataset from 1985-2010 (shown in Figure 2).
Figure 1: SST anomaly of December 2013, January and February 2014 compared to the 1985-2010 OSTIA DJF climatology
Figure 2: DJF climatology for the period 1985-2010 in the OSTIA dataset for sea surface temperatures
Similarly to Sparrow et al. (2013), the “current winter in the world that might have been” experiments are obtained by removing the modelled SST patterns of anthropogenic forcing from the observed 2013/2014 SSTs. By doing so, we are simulating the winter 2013/2014 in a counterfactual world where there was no human influence. The modelled SST patterns resulting from the anthropogenic forcings are calculated from the CMIP5 archive. This project ran two climate model experiments: one, “Historical” included both human-caused greenhouse gas emissions and natural emissions, such as volcanoes; the second, “HistorialNat” included only the natural emissions, and deliberately left out human-caused emissions, to see how the climate might have changed without them. We selected the models that have more than 3 ensemble members for both the Historical and HistoricalNat experiments.
For each model and experiment, the monthly climatologies have been calculated for the period 1996-2005 and averaged over all ensemble members available. 1996-2005 is the last decade simulated as part of the historical experiments of CMIP5. Then we subtracted the monthly climatologies of HistoricalNat from the corresponding ones in Historical. These anomaly patterns represent the effect of the anthropogenic forcing on the SSTs. Figure 3 shows this pattern for the multi-model mean (MMM) and for DJF. The pattern shows an overall warming, but in individual models, this can be relatively different. Figure 4 shows the two most different patterns we obtained: on the left, a relatively cold one, and on the right, a relatively warm one.
Figure 3: DJF SST response pattern to anthropogenic forcing averaged over all CMIP5 members available
Figure 4: DJF SST response pattern to anthropogenic forcing for the GFDL-CM3 model (left) and the CCSM4 model (right).
By removing these anthropogenic patterns from the observed 2013/2014 SSTs, we expect to obtain a plausible range of possible outcomes given that there are substantial variations in these patterns across the models. We are also removing the anthropogenic pattern of the MMM, which might provide a clearer response as the pattern is by definition smoother than in the individual models.