What can high-resolution regional climate simulations tell us about the future urban climate of European cities?

Extreme weather events such as heat waves and heavy precipitation are already having an impact on urban areas around the world, and their frequency and/or intensity are expected to increase as a result of ongoing global warming. Local decision-makers need high-resolution urban climate information to plan and adapt tomorrow's cities. This information needs to be tailored to their needs and to different geographical contexts (e.g. by representing mountainous areas, coastal lines or city characteristics). It must also be appropriate in terms of time scale (e.g. from a particular extreme event to climatological timescales) and cover the range of uncertainties. Today, regional climate information is often derived from Global Climate Models (GCMs) that are downscaled to the local scale using statistical tools, or Regional Climate Models (RCMs) such as those used in the CORDEX initiative. Long-term RCM simulations reach horizontal resolutions of the order of ten kilometers and offer added value in certain respects compared with their driving GCM, but these resolutions are not sufficient in certain specific contexts such as cities, or in highly heterogeneous mountainous areas or along coastlines. The latest generation of RCMs, known as Convection Permitting Regional Climate Models (CPRCMs), now achieve resolution down to the kilometer scale and can be used to better represent these heterogeneous land surfaces, potentially offering new insights into local climate change. Here, we analyze simulations carried out as part of the CORDEX Flagship Pilot Study on Convection (3 km). We first investigate the CPRCMs ability to represent the historical urban climate of different European cities, compared with their lower-resolution counterparts. We then study the evolution of various urban climate impact indicators in the future under a high-emissions scenario. We analyze the effects of the increased resolution, choice of urban parameterizations, land cover representation approaches (dominant versus fractional cover approaches) and land cover datasets. Finally, we compare the range of uncertainties displayed by the new high-resolution simulations against the previous CORDEX ensemble.