The climate is variable on all timescales. A key challenge remains the prediction of changes in climate mean, variability and extremes (high-impact weather and climate events) for seasons, decades and centuries ahead. On shorter time scales, operational probabilistic climate predictions, mostly based on dynamical models, have been implemented in a number of forecasting institutions and practical applications are being developed. On seasonal-to-decadal and longer time scales, recent developments in modelling and statistical (downscaling) methods within projects like SPECS, EUPORIAS, CORDEX, COST-VALUE or the IPCC AR5 context provide the base to develop refined local, regional and global climate predictions and projections using ensemble and probabilistic techniques. Aside from the scientific hurdles, a key challenge remains in the tailoring of such climate predictions and scenarios to cover end user needs and climate change impacts assessments.

The session invites papers, particularly those focusing on ensemble approaches, related to:

• Seasonal-to-decadal climate forecasts (results from the SPECS and EUPORIAS FP7 projects, predictability, multi-models, verification, calibration, downscaling, extremes, statistically based schemes, tailoring to end user needs,..).

• Local, regional and global climate change scenarios (global vs. regional models, CMIP5, CORDEX developments, multi-model approaches, post-processing, bias correction, statistical downscaling, quantifying model uncertainties, quantifying changes in extremes, tailoring to end user needs, ..).

• Seamless predictions: approaches to bridging the gap between weather and climate forecasts.

• Tailoring climate forecasts and longer-term projections for impacts assessments and use by societal sectors (statistical downscaling techniques, e.g. developed within COST-VALUE).