The IPSL “Earth System model” is currently composed of a climate model and a runtime environment (libIGCM) handling the production of simulations (both model execution tasks and output data post-processing tasks) on the IDRIS and TGCC computer centres. The platform will be built based on libIGCM that will be progressively improved by including all the tools developed within this project. This progressive approach has been chosen to make sure that the tools developed match their objectives and are operational and effective in real cases. The development of the platform will be organized in two phases over a four years period:

  1. Within two years, a first version of the whole platform will be realized and tested through an implementation of a standard version of the IPSL-CM model. The main objective of this step will be to ensure the overall consistency of all the tools developed within the project and to have a first feedback of their strength and weaknesses.
  2. Within the two last years of the project, the developments will follow two directions. Firstly, all the tools of the platform will continue to evolve to reach their final objectives. Secondly, the platform will be used to implement three new climate model configurations.

The project will be organized around six main tasks.

  • Task 0 concerns the management of the project, the valorisation and the sustainability of platform.
  • Task 1 will ensure the platform release, the documentation and training. It will realize four typical implementations of the platform and an example of its challenging use.
  • Task 2 aims at developing and providing all the necessary tools and environments to allow running high resolution coupled models. It includes mixed MPI/OpenMP parallelism optimisation, managing IO and scale aware parametrizations, parallel interpolation and coupling mechanisms.
  • Task 3 aims at the ability to manage all aspects of high-performance scientific workflows. The goal is to connect our family of models to state-of-the-art HPC computing platform and to develop a supervisor daemon to ensure seamless completion of long lasting high end climate simulations.
  • Task 4 deals with Big Data challenges. The goal is to provide the data management and analytics component in order to ensure continual delivery of a comprehensive end-to-end solution able to bring together large volumes of diverse data at any velocity to generate new insights.
  • Task 5 deals with evaluation of climate simulations data. We propose a generic environment leveraging the data services infrastructure deployed in task four. Taking for input simulation results and observational data, this task will provide a set of pertinent analysis components targeting a framework for seamless evaluation of the climate models.

convergence tasks