The main objectives of ESiWACE2 are:

(1) Enabling leading European weather and climate models to leverage the available performance of pre-exascale systems with regard to both compute and data capacity in 2021.

(2) Preparing the weather and climate community to be able to make use of exascale systems when they become available.

To achieve these high-level objectives, the ESiWACE2 consortium has set a number of specific objectives, which lay the foundation for the work plan of this proposal.

• Boosting European climate and weather models to operate in world-leading quality on existing supercomputing and future pre-exascale platforms.
• Establishing new technologies for weather and climate modelling.
• Enhancing HPC capacity of the weather and climate community.
• Improving the toolchain to manage data from climate and weather simulations at scale.
• Strengthening the interaction with the European HPC ecosystem.
• Fostering co-design between model developers, HPC manufacturers and HPC centers.

CMCC ROLE

In ESiWACE2, CMCC will lead WP5 on Data Post-Processing, Analytics and Visualization. The main objective of WP5 is to provide a consistent view regarding the support for data postprocessing, analytics and visualisation at scale in the weather and climate domain by building, on top of the ESDM module developed by WP4, the proper ESDM extensions.

In particular, the tasks in WP5 concern a) the design of the ESDM interface extensions to support in-flight analytics kernels for post-processing, analysis and visualization (PAV) needs; b) the identification, prioritization, implementation and validation of a set of common analytical kernels starting from a set of community-based tools; c) the development of a high-performance support to enable ESDM data parallelization for in-flight analytics; d) the validation of the ESDM postprocessing, analytics and visualization support on a set of representative case studies regarding community-based weather and climate applications.

More specifically, CMCC will also contribute to:

a) WP1, regarding the development of the infrastructure for ocean production mode configurations and the porting to pre-exascale systems;

b) WP2, by leading the task on watching emerging technologies and by contributing to the application of DSLs to ocean models;

c) WP4, by contributing to the ESDM implementation and testing activities;

d) WP5, by leading the WP and contributing to the design and implementation of the ESDM PAV runtime, as well as its interface to post-processing and analytics applications;

e) WP6, by leading the training on HPDA tools and frameworks.

ACTIVITIES

The work plan of the project is arranged in 6 work packages, aimed at:

• improving the throughput and scalability of leading European weather and climate models and demonstrating the technical and scientific performance of the models in unprecedented resolution on preexascale EuroHPC systems (WP1).
• developing and establishing domain- specific languages (DSLs) for community models and evaluating the potential of innovative technologies such as machine learning to explore new methodologies in Earth system modelling (WP2).
• providing support and training to the community to profile and optimize code performance, and to port models to existing and upcoming exascale supercomputers (WP3, WP6).
• developing tools to mitigate the effects of the data deluge from high-resolution simulation: ensemble tools to minimize data output, storage middleware to handle IO performance, and improved tools for post-processing data analytics and visualization. (WP4, WP5)
• reaching out to the European HPC initiatives and, specifically, the related FET projects, PRACE and EuroHPC, and providing platforms for dedicated interactions between the climate and weather community and HPC stakeholders (WP6).
• co-design, that will be supported through the High-Performance Climate and Weather benchmark (HPCW; WP3) but also by all of other WPs.

EXPECTED RESULTS

The vision of ESiWACE2 is to take a significant step towards production-ready weather and climate models at (pre-) exascale based on existing software on the one hand, and to significantly advance weather and climate simulation technology as well as the community as a whole—in terms of HPC-awareness and HPC capabilities—on the other hand.

The ambition of the project is to establish domain-specific languages in Earth system modelling and to transform the approach to data handling, fostering co-design and the exchange of model codes, exploiting software-defined storage, evaluating DSL approaches in various models, establishing code optimization techniques and sharing knowhow from one group to another, in order to overcome the performance barriers for Earth system models.

The expected impacts are:

• European leadership in exascale and extreme-scale-oriented codes and innovative algorithms/solutions that address societal challenges or are important for key scientific and industrial applications
• Improved access to computing applications and expertise that enables researchers and industry to be more productive, leading to scientific excellence and economic and social benefit
• Improved competitiveness for European companies and SMEs through access to CoE expertise and services
• Federating capabilities and integrating communities around computational science in Europe
• A large number of scientists and engineers, in particular female and young ones, trained in the use of computational methods and optimization of applications.

PARTNERS

• CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE CNRS (CNRS-IPSL)

• EUROPEAN CENTRE FOR MEDIUM-RANGE WEATHER FORECASTS (ECMWF)
• BARCELONA SUPERCOMPUTING CENTER (BSC)
• MAX-PLANCK-GESELLSCHAFT ZUR FORDERUNG DER WISSENSCHAFTEN EV (MPIM)
• SVERIGES METEOROLOGISKA OCH HYDROLOGISKA INSTITUT (SMHI)
• CENTRE EUROPEEN DE RECHERCHE ET DE FORMATION AVANCEE EN CALCUL SCIENTIFIQUE (CERFACS)
• NATIONAL UNIVERSITY OF IRELAND GALWAY (ICHEC)
• MET OFFICE (METO)
• THE UNIVERSITY OF READING (UREAD)
• UNITED KINGDOM RESEARCH AND INNOVATION (UKRI)
• BULL SAS (BULL)
• SEAGATE SYSTEMS UK LIMITED (SEAGATE)
• EIDGENOESSISCHE TECHNISCHE HOCHSCHULE ZUERICH (ETH Zürich)
• THE UNIVERSITY OF MANCHESTER (UNIMAN)
• STICHTING NETHERLANDS ESCIENCE CENTER (NLeSC)
• EIDGENOESSISCHES DEPARTEMENT DES INNERN (MeteoSwiss)
• DATADIRECT NETWORKS FRANCE (DDN)
• MERCATOR OCEAN (MO)