Seasonal forecasts and heat stress in an eastern Mediterranean environment

In this study we examine the use of seasonal forecasts on predicting the heat stress conditions in the summer months over the region of Attica in Greece using a number of meteorologically based heat stress indices, namely effective temperature, humidex, discomfort index and heat index (Buzan et al. 2015). To this aim, the fifth generation ECMWF seasonal forecasting system (SEAS5) hindcasts for the period 1993 to 2016 available in C3S Climate Data Store are used. The variables to calculate daily heat stress indices values include instantaneous outputs at 12 UTC for 2-meter temperature, northward and eastward near-surface wind components, as well as 2-m dewpoint temperature. In order to statistically downscale and verify seasonal forecasts, the state-of-the-art global reanalysis dataset ERA5-Land of Copernicus CDS is used. The verification of the heat stress indices re-forecasts is performed using adequate probabilistic verification measures of discrimination and reliability.

Preliminary results indicate that for the majority of the indices, forecasts initialized in June and to a lesser extent in May exhibit statistically significant skill scores in predicting above normal heat stress conditions, while for forecasts initialized in March and April no statistically significant skill is found identifying these forecasts in the not useful category.