GMGEC group
Team Leader : Damien Specq
Field of activity
The PASTEL team (Predictability of the Atmosphere at Seasonal timescales and TELeconnections) focuses on research in numerical predictability at monthly to interannual timescales, with a particular emphasis on the seasonal scale. To achieve this, the team uses the CNRM-CM global coupled climate system model. It contributes to model improvement by studying sensitivity to resolution and physics at this scale. Within an international framework, the team robustly evaluates the quality of forecasts, as well as the sources of predictability (ocean, land surface, teleconnections).
The team is also responsible for producing forecasts up to seven months ahead, thereby contributing to the climate services provided by Météo-France.
Research topics
The main research themes of the PASTEL team are:
Understanding sources of predictability and teleconnections
This activity involves studying the various sources of climate predictability, with the aim of evaluating and improving their representation in coupled climate model-based forecasting systems. These sources include the slow components of the climate system (ocean, land surfaces, etc.), as well as statistical connections—whether chronological or not—between a predictable phenomenon (such as ENSO, QBO, etc.) and one that is less so. It also involves conducting experiments in a more idealized framework and analyzing observational data.
Designing and operating forecasting systems
This activity focuses on the scientific challenges associated with the development of climate forecasting tools in a real-time context. Research addresses the methods used to produce these forecasts, whether through physical models (initialization strategies, ensemble generation, resolution, and parameterizations) or statistical approaches (machine learning and artificial intelligence). This work also includes operational production for seasonal forecasts (Copernicus Climate Change Service, C3S) and semi-operational production for sub-seasonal forecasts (Subseasonal-to-Seasonal, S2S collaboration).
Evaluating and interpreting seasonal and monthly forecasts
This activity encompasses all research conducted on forecast data after their production. It includes verifying forecasts against observational data and improving them a posteriori through post-processing. It also leverages the ensemble forecasts to explore the range of possible climate states at seasonal and monthly lead times, with several objectives: quantifying uncertainties, identifying the most realistic forecasts a priori, and assessing the probability of extreme events.
Forecasting systems
In addition to its research activities, the PASTEL team carries out engineering work to develop, maintain, and operate two forecasting systems:
- Météo-France’s operational seasonal forecasting system, which produces real-time forecast data on a monthly basis. This data is provided under contract to the European Copernicus Climate Change Service (C3S).
- A monthly forecasting system (up to 46 days lead time), which generates forecasts every week to feed into the database of the international Subseasonal-to-Seasonal (S2S) initiative.