AIT explores and applies AI and machine learning methods to solve Earth Sciences problems, such as speeding up numerical simulations, improving predictive skill, and creating human-machine interfaces. Additionally, AIT supports data-driven research in climate and atmospheric sciences.

See full list here

Develops and applies Machine Learning and Artificial Intelligence methods to address key challenges in Earth sciences in the department.

• Design, develop or adapt AI models related to Earth Sciences
• Collaborate with domain scientists
• Build and maintain AI-ready datasets and benchmarks
• Support reproducible and efficient AI research
• Promote knowledge transfer and outreach
• Contribute to reserach in AI for Earth Sciences

AI Models:

• LLM for climate information (under development)
• ClimateDT emulator (under development)
• LAI, LU, LC classification (under development)
• eMONARCH

Projects:

• LLM4ES (DE_340)
• ClimateDT emulator (DE_393)
• AI4Land (CONCERTO, TerraDT, ELLIOT)
• eMONARCH (Fundos de recuperacion)
• Open Science Platform (EXPECT)

See full list here

Maintain the ES department infrastructure (data storage and software stack) and develop data analysis tools and pipelines for national and international projects.

• Develop, manage and maintain a common data service framework.
  • Develop, manage and maintain a common data service framework to collect, standardize and distribute climate and atmospheric data to both internal departmental and external users of the research and services community. Setting up an ESGF Data-Node is a priority of the team to give international visibility to the centre.
  • Implement locally and contribute to the development of international standards for data storage and exchange, with special interest in data in NetCDF and GRIB formats.
  • Contribute to external data quality checking projects (eg Copernicus C3S_512)

• Deploy an infrastructure ready to overcome the Big Data challenge in Earth sciences.
  • Improve the capability for processing Big Data volumes for the analysis of Earth system simulation using the latest technologies both at hardware and software levels.
  • Improve the outcome to society of user-friendly data visualization products.
  • Study new approaches to address the I/O challenges of the new generation of high-resolution, highly parallelized Earth system models.
  • In this page, you can find the current material related to Big Data in Earth Sciences.

• Artificial Intelligence for Earth Sciences.
  • New developments on Machine Learning and Artificial Intelligence have great potential for the exploitation and modelling of the ever-increasing stream of geospatial data. Therefore, a new research line has been started within CES to develop new Machine Learning models (specially based on Deep Neural Networks) for Earth Sciences. All the information can be found here (AI4ES).

Projects:

• DestinationEarth
• EERIE
• ESGF
• RESPIRE
• CALIOPE
• ACTRIS
• BDRC
• Copernicus
• IDAlert
• HARMONIZE

Softwares:

• R packages from the R unit
• EC-Earth postprocessing
• Providentia
• Dashboards Caliope, BDRC, Epioutlook

Others:

• C3PO
• Earth Diagnostics
• ece2cmor
• R2D2
• SDS-WAS/BDFC Map Viewer and Other customizations
• MapGenerator
• ESGF Node: info and esgf-ceda portal
• THREDDS info and THREDDS direct link
• s2dverification
• C3S_512

Efficient of use of the data by users and meet the users needs for data analysis.

See full list here

Efficient use of the computational resources by the research groups

• Provide HPC Services such as performance analysis to identify bottlenecks and apply optimizations
  • Collaborate with other BSC departments, especially Computer Sciences, to use state-of-the-art programming models and profiling tools to prepare Earth sciences models to run on next generation exascale HPC system. Also improve the efficiency of existing models and postprocessing tools interacting with developers and users during all stages of software development lifecycle.

• Research on new computational methods to apply on Earth Sciences models
  • Collaborate with external Earth system modelling teams to implement and test new computational methods that allow the use of new platforms such as heterogeneous architectures.

• ELPin
• Lucia-lite

See full list here

Development of HPC user-friendly software framework for Earth system modeling and the management of operational systems

• Support the development of atmospheric research software and contribute to its maintenance.
  • Interact with model developers and HPC support teams to develop and deploy a software stack to run generic Earth system models on a wide range of HPC facilities.

• Maintain and improve operational systems.
  • Collaborate with the other three groups to satisfy their needs for the development and efficient running, including the design of appropriate workflows, of the BSC-ES operational air quality, weather and climate forecast systems.