An outstanding question for climate science is quantifying how global warming will regionally affect the aspects of climate that are most directly relevant to society, such as precipitation, windiness and extremes. But achieving this task is proving not to be simple. The main available tool consists in computer simulations performed using ensembles of climate models. These models are used to run scenarios in which greenhouse gas concentrations increase with time, so that the climate response to warming can be evaluated. But, for the aspects of climate that are controlled by the atmospheric circulation, there still remains substantial spread across the model projections thus leading to uncertainties in how regional climate will respond to global warming.
Figure 1:Decadal evolution of cold season (November to April) Mediterranean precipitation as a function of global warming in the simulations from 36 climate models from the 5th phase of the Coupled Model Inter-comparison Project (CMIP5) for the RCP8.5 emissions scenario. Precipitation and temperature changes are evaluated relative to the 1960 to 1990 mean. The thick line shows the multi-model. For presentation purposes, the horizontal axis starts at 0.2 K.
Let’s consider Mediterranean precipitation as an example. Figure 1 shows that while the average of 36 climate model projections, as well as most of the individual models, indicate a future decline in winter Mediterranean precipitation, the magnitude of the precipitation reduction, even for a given warming of the planet, remains highly uncertain. Notably, the projected drying at 2 degrees warming in some models can be larger than the drying at 4 degrees warming in other models. Taking the multi-model mean provides a simple, and often adopted, approach to summarise the ensemble and communicate the regional projections to stakeholders and decision makers. But substantial information on the uncertainty is lost by simply averaging the model responses. So is this fully justified?
In a recent paper, Zappa and Shepherd propose to use an alternative storyline approach to characterise the uncertainty in regional climate projections from ensembles of climate models. To think in terms of storylines, it is necessary to realise that regional atmospheric circulation can be driven by remote aspects of climate. This is true on seasonal timescales, for example in response to the development of El Nino or La Nina events in the tropical Pacific, but it is equally true for the long timescales associated with climate change. In particular, Zappa and Shepherd identify two remote drivers of atmospheric circulation whose response to climate change is both uncertain and capable of influencing the European and Mediterranean climate: the magnitude of the upper tropospheric warming in the tropics and the strength of the Northern Hemisphere stratospheric vortex.
Figure 2:Four different plausible storylines of cold season Mediterranean precipitation change per degree of global warming. The different storylines depend on the magnitude of the tropical amplification of global warming and on the strength of the stratospheric vortex response as indicated above the figures. See Zappa and Shepherd 2017 for more details.
By applying a statistical framework to the climate models output, four different plausible storylines of Mediterranean precipitation change are identified for different combinations in the two remote drivers responses (see Figure 2). The patterns of regional precipitation change per degree of global warming within each storyline are found to be rather diverse: depending on the storyline, the Mediterranean precipitation response can be larger (Figure 2b) or weaker (Figure 2c), or it can be more focused on the eastern (Figure 2d) or on the western (Figure 2a) Mediterranean. A worst case storyline of Mediterranean climate change is identified for a large tropical amplification of global warming and a strengthening of the stratospheric vortex (Figure 2b), in which case the Mediterranean drying per degree of global warming is expected to be particularly enhanced.
Until there is sufficient physical understanding or observational evidence to discard one of the above combination of driver responses, these four storylines should be considered equally plausible future realisations of Mediterranean regional climate change. It is worth to highlight that, until discarded, the worst case storyline could still be realised. This should be kept in mind when evaluating the risks of climate change and developing local adaptation plans.