Seasonal-to-decadal climate Prediction for the improvement of European Climate Services

There is an increasing need in society for robust climate information covering future periods ranging from several months up to several years for economic, industrial and political planning. Prediction on this near-term climate time scale has received less attention than weather forecasting and long-term climate change projection. Seasonal-to-decadal predictions have currently limited forecast quality, and improving atmospheric forecast skill beyond the synoptic time scale requires further exploiting of the memory inherent to the slowly varying components of the climate system. In particular, recent developments indicate that the sea ice, land surface, stratosphere and ocean components might allow taking advantage of untapped climate predictability. There has been renewed interest in tapping on that potential with state-of-the-art, dynamical prediction systems. This project is linked to the SPECS EU-project (http://www.specs-fp7.eu/) aimed at developing a new generation of European climate forecast systems. Innovative global forecast experiments on seasonal time scale show that a moderate forecast skill improvement can be obtained with better initialisation of the ocean and the cryosphere. In particular, we have demonstrated that a realistic snow initialisation improves forecasts at high northern latitudes and in the Arctic. We have also compared the quality of snow data in various meteorological and climate analyses used to initialise forecasts. This effort has allowed us to initiate and lead a science project for the World Climate Research Programme on the role of snow in seasonal forecasting. In collaboration with scientists at the European Centre for Medium-Range Weather Forecasting in England, we carried out and analysed the first simulations in that project, and determined that snow initialisation impacts surface temperature over parts of Eurasia at the monthly time scale. Also, the snow prediction itself is more reliable with realistic snow initialisation. The operational seasonal forecasts of the European Centre for Medium-Range Weather Forecasting also show two categories of stratospheric disturbances, depending on their duration (short or long-lived). While short disturbances occur near-simultaneously throughout the stratosphere, only the long-duration ones exert a distinct influence on surface weather.

There is increasing need for robust climate information covering future periods ranging from several months up to several years for economic, industrial and political planning. Prediction on this near-term climate time scale has received less attention th an weather forecasting and long-term climate change projection. Seasonal-to-decadal predictions have currently limited forecast quality, but recent developments indicate that sea ice, land surface, stratosphere and ocean that might allow taking advantage of untapped climate predictability. Initialisation is one of the critical stages in dynamical prediction. The potential impact on European climate predictability such as the initialisation of the land surface, snow cover or sea ice, have received incre ased attention up to now. In this project, we will further explore potential sources of predictability on the seasonal time scale, arising from cryosphere-atmosphere interactions (in particular snow), and from the polar middle atmosphere, with a special focus on the high northern latitudes and the Arctic.