Many European countries have designed weather early warning systems of heat stress indicators. This process was especially evident after the summer 2003 heatwave, which had an unprecedented impact on mortality across the continent, causing more than 70,000 premature deaths in western countries alone. Implementing adequate health preventing measures, which have a positive impact on reducing temperature-attributable mortality (TAM), is essential in public health decision making, particularly in a context of climate change and rising temperatures. Yet, these systems have room for improvement. We identified some of the key aspects that could be refined after consulting several stakeholders and end-users from a wide-range of professional fields. They emphasised the need of a unified Pan-European service that provides relevant and comparable information. At the same time, the service should be flexible enough to adapt to the different climatic and socio-economic structures of the European societies. Since most of the current systems are solely based on climate data, they also highlighted the need to include mortality data to model the real impact of weather and climate. Lastly, end-users demanded that the system was able to produce warnings for multiple lead times beyond the traditionally used 1 to 2 days. Hence, to design prototypes of European weather early warning systems that address these needs, our objective was to study the predictability of temperature-attributable mortality in Europe at the regional scale using weather forecasts with lead times of up to 15 days. Here we present the results of our study. Our main finding is that temperature predictability can be transformed into TAM predictability. Due to the differences in the temperature-mortality associations, significant differences in the TAM predictability are found across the regions. These differences would not be identified if only temperature forecasts were considered. We have seen better predictability in summer for regions associated with a high risk of mortality for summer temperatures, such as the Mediterranean and the northern regions of Germany and the Netherlands. While for winter, better skill is found in regions with a different temperature-mortality association; for example in France, Poland and the United Kingdom. Although it depends on the region and season, in general TAM can be predicted on weather time scales, as lead times with useful skill are comparable after the transformation of temperature into TAM. In addition, there is a relationship between temperature predictability and TAM predictability, so we presume that future improvement in the weather forecasting will directly lead to improvements in TAM forecasting.

The Blue-Action project has received funding from the European Union's Horizon 2020 Research and Innovation Programme under Grant Agreement No 727852.