The impacts of a global warming of 1.5°C, and the impacts avoided by stabilizing temperatures at 1.5°C instead of 2°C, will be dominated in many regions by changing risks of extreme weather events exceeding critical thresholds. Relatively small ensembles of coupled model integrations driven by specified future emissions, as requested by CMIP, are primarily suited to the assessment of expected changes in mean climate within an emissions-based scenario, not weather extremes within a temperature-based scenario. Quantifying these changes requires large initial-condition ensembles of high resolution atmospheric models driven under temperature-based scenarios.

The attribution community has been using large ensembles to deal with low signal-to- noise problems for over a decade, and their methodology could be directly applied to this climate projection problem. To directly address impact differences between the current and 1.5°C and 2°C worlds, the HAPPI project is generating large ensembles (>50 members) of ten-year periods for recent observed, and 1.5°C and 2°C warmer worlds, using projected changes in sea surface temperatures drawn from existing coupled-model simulations. The
use of ten-year time slices allows for the assessment of long-lived extreme events, such as droughts, while still allowing for large ensembles. The resultant probabilistic assessment of climate allows for precise quantification of changes in extreme weather given ambitiously small changes in global temperature.

Here, we present initial results from the HAPPI project, linking changes in climate extremes through to changes in a range of impacts.