Research

CHAPTERS: [Climate] [Couplers] [Data assimilation]

Climate

The "climate group" is organized into three projects :
- Climate variability and predictability
- Climate change and related impacts
- High resolution climate models

The various projects have deep interaction with each other through the existence of specific transverse activities (research projects, Phd thesis, contracts, formation activities ...) and also with the other groups (oceanic data assimilation, OASIS, PALM) within the team.

The global coupled atmosphere-ocean-land-cryosphere system exhibits a wide range of physical and dynamical phenomena with associated physical, biological and chemical feedbacks that collectively result in a continuum of temporal and spatial scales and variability. In addition to this intrinsic variability, the world climate is currently changing due to the human-induced increase in the atmospheric concentrations of CO2 and other greenhouse gases. Our current limited understanding of these complex multiscale interactions and and therefore the ability to advance weather and climate prediction is at the core of climate research. A new cross-cutting domain, decadal prediction, is slowly emerging at the interface between short (weather) and long (climate) time scales. It mixes predictability of the rst (initial conditions) and second (external forcing) kind and may appear as an area of methodological convergence of weather and climate forecasting. It has also a growing importance in terms of the adaptation and mitigation strategies needed to cope with the potential impacts of future climate change at both global and regional scales.

The "climate" group conducts basic and applied research in the eld of climate studies. Our long-time main objective is to make signi cant contributions to the understanding of the world's climate variability on regional to global scales with the aim to improve climate forecasts as well as climate change impact studies at seasonal-to-decadal time scales. We are currently beginning a new research cycle on decadal predictability issues, with the first objective to actively participate to the next IPCC exercise. The question of the existence of significant potential predictability at decadal time scale is an open research issue. The time scale of natural decadal climate variations overlaps with expected trends associated with anthropogenic climate warming, thus presenting an important signal-to-noise problem for attributing observed trends over the previous and next few decades. Understanding the mechanisms of decadal climate variations, quantifying predictability, coordinating the required observing system, and developing prediction systems (i.e., enhancing modeling capabilities, initialization strategies and effective use of observational estimates), are key goals at the core of the climate research community.

Our own and immediate strategy concerns the development of a low-resolution (1 degree) decadal prediction system with a strong focus on the question of system initialization and bias reduction as well as understanding of the mechanisms underlying the climate variations at decadal scale. This activity bene ts very strongly from a collaboration with the CNRM. In parallel, we have also started to investigate short-term climate simulations using high-resolution coupled modelling. Encouraged by the emergence of very large-scale computing platforms and in order to better simulate regional climate features and climate sensitivities, our long-term grand challenge (10 years) is to develop a decadal prediction system with improved physics and a signi cant increase in resolution. We envision a system based on the coupling of an atmosphere and land model with either a 25 km (0.25 degree) or 50 km (0.5 degree) separation between grid points to an ocean and sea ice model with 10 km (0.1 degree) separation. This grand challenge raises a large number of scienti c and technical problems that we aim to address in the coming years. The full implementation plan for this grand challenge will be discussed with the interested CERFACS partners in 2010 and 2011. We will also continue our backbone activities regarding the development of high-level engineering softwares needed to address the high-resolution related issues arising in climate science (effciency and scalability of coupling software, data compression and reduction, fault tolerance).

Couplers

The "Coupler" group of projects gathers two items:
- The OASIS coupler and its applications
- The PALM software

Coupling numerical models, i.e. implementing synchronized exchanges of information between  models, is a central issue in many research elds such as climate modelling, data assimilation,  or computational  uid dynamics. Since about 20 years, CERFACS specializes in developing  coupling software tools, also called "couplers".

Today, about 30 climate modelling groups in the world are actively using the OASIS coupler,  either the robust OASIS3 version or the new fully parallel OASIS4 version. The PALM  coupler, addressing parallel dynamic coupling, is used by about 30 research and development  groups in France for multiphysics purposes. Several CERFACS partners (Météo-France, EDF,  SNECMA, ONERA, TOTAL, ...) are using one of these of two couplers and priority is given to  developments which match their needs. These activities allow CERFACS to play a central role  in the adaptation of coupled applications to the new and future massively parallel platforms.

Data Assimilation

The  Data Assimilation" group of projects the following items :
- Ocean data assimilation
- Data assimilation for nuclear core
- Data assimilation for hydraulics and hydrology

The  "Ocean data assimilation" project is based on the development of the NEMOVAR system through an international collaboration. The "Data assimilation for nuclear core" project includes strong collaboration with the EDF/SINETICS Department.. The Data assimilation for hydraulics and hydrology" project includes strong collaboration with the EDF/LNHE Department.
 
In addition to these specialized topics, sustained e orts will continue to disseminate the pratice of data assimilation in a growing number of domains. Several research or development teams, often coming from CERFACS partners, come and visit CERFACS to gain knowledge about data assimilation. It is planned to develop advanced traning on the topics, beyond the PALM and PALM/Assimilation three day training periods which already exist and will continue. New training on data assimilation will be developed in collaboration with CERFACS partners, principally with EDF/R&D with whom a rst experience in this direction has already been performed.