Influence of trends on subseasonal temperature prediction skill

Subseasonal-to-seasonal (S2S) predictions have a wide range of applications. Improving forecasts on this time-scale has therefore become major effort. To evaluate their performance, these are routinely compared to reference that the climatological distribution at any given time. This is commonly assumed be stationary over verification period time-scales longer than seasonal cycle. However, there prominent deviations from assumption, especially considering trends associated with climate change. By employing synthetic forecast-verification pairs, we show estimates probabilistic skill both continuous and categorical increase as function variance explained by trend period, even when errors in simulated forecasts. We also enhancement due ECMWF extended-range ensemble prediction system. demonstrate effects an operational forecast setting currently strongest Tropics. Our results care needs taken evaluating subject non-stationarity much window. important for determining forecasts, where assumptions stationarity climatology enter determination category thresholds. The presented study not exclusive S2S but wider implications decadal time-scales, existence can further impact skill.