Co-Leader, Computational Earth Sciences at @BSC_CNS | ESiWACE Coordinator — a Horizon Europe/EuroHPC Centre of Excellence for weather & climate simulation
Barcelona, Spain
Joined July 2018
- Tweets109
- Following66
- Followers59
- Likes152
🌍 Yesterday I chaired HANAMI’s climate session in Kittilä. From model evaluation to Autosubmit and One-Pass, our BSC team joined European and Japanese colleagues to connect climate science and HPC.
Explore the programme: hanami-project.com/2026/01/1…
🤖 Made with AI
🌍 Eight teams, nine mentors, one shared challenge: preparing weather and climate models for JUPITER. Our ESiWACE hackathon brought scientists and HPC experts together to work on GPU porting, scaling and workflows.
A look back: esiwace.eu/news/e-si-wace-3-…
AI weather models also need well-organised experiments. In our CES group at BSC, AIUQ connects AI-based weather and climate forecasting models with Autosubmit, with flexible handling of initial conditions. Explore the public code: gitlab.earth.bsc.es/ces/AIUQ
Looking to sharpen your HPC skills for weather and climate modelling? In ESiWACE, which I coordinate at BSC, our partners share free training materials on profiling, GPUs and more. Explore and share: esiwace.eu/training-event/
We’re hiring! New PhD student position in the CES group at BSC, funded by a Spanish Ministry FPI grant. Research HPC performance, efficient weather & climate workflows and AI agents with us in Barcelona. Apply by 19 Nov:
bsc.es/join-us/fellowships/4…
PDP 2027 is coming to Barcelona on 17–19 March, bringing together work on parallel and distributed computing, HPC, scientific workflows and AI at scale. Abstract registration closes 19 October. Glad to support the conference as a Program Committee member. research.ac.upc.edu/PDP2027/…
Two project kick-offs, one shared goal: preparing European climate modelling for the next generation of supercomputers. In Milan I presented ESiWACE4 to ENES-RISe and FUTURA, building the connections we’ll need before the project starts in 2027. esiwace.eu/news/e-si-wace-pr…
Climate workflows can lose days waiting in supercomputer queues. With Pablo Goitia, Manuel Marciani and Miguel Castrillo, we found that Autosubmit task aggregation reduced total queue time 11–12-fold across three European systems. doi.org/10.5194/gmd-19-8501-…
New Climate DT data is now available: our Computational Earth Sciences team at BSC developed the Energy Indicators application, turning 5 km, hourly simulations into global metrics for wind and solar potential, energy demand and extremes. destine.ecmwf.int/news/clima…
Today we kick off CLAIMA, a new Horizon Europe project where our Computational Earth Sciences team at BSC will help build a trustworthy AI-driven climate digital twin—linking emulators, downscaling and foundation models with EuroHPC and DestinE. cordis.europa.eu/project/id/…
How can kilometre-scale climate simulations support adaptation? Our paper, led by Paco Doblas-Reyes, presents the Destination Earth Climate DT: ICON, IFS-FESOM and IFS-NEMO working within one operational framework. doi.org/10.5194/gmd-19-2821-…
Climate simulations involve thousands of tasks on HPC systems. Our CES team at @BSC_CNS develops Autosubmit to orchestrate reproducible workflows, including ClimateDT runs with IFS-NEMO, IFS-FESOM and ICON. Scientific software is infrastructure. github.com/BSC-ES/autosubmit
What does Computational Earth Sciences look like in practice? In our CES group at @BSC_CNS, 90+ people connect four areas: AI, data & diagnostics, HPC, and models & workflows. Different expertise, one goal: making Earth-system science more capable, efficient and reproducible.
Scientific leadership is also about creating spaces where communities meet. On 15 November at #SC26, I’ll help organise HPC4EES in Chicago, connecting HPC, AI and Earth sciences—from climate modelling to hazards and energy efficiency. hpc4ees.github.io/
Glad to have contributed as co-author to our BSC team’s recent MareNostrum 5 paper. We go beyond peak figures, evaluating architecture, scalability and energy efficiency with workloads—including IFS for weather and climate. doi.org/10.1016/j.future.202…
Climate digital twins are more than larger models. In our CES/BSC work within TerraDT, science, software, workflows and supercomputers must evolve together. Co-design is not a project phase; it is the operating model for km-scale climate systems. terradt.eu/
Europe’s climate-modelling community needs more than faster machines: it needs reliable services linking models, data and evaluation. ENES-RISe brings 22 partners together to strengthen that shared infrastructure and make climate projections easier to access and use.
AI can speed up weather and climate information—but it can also deepen inequalities in compute, data and whose knowledge counts. Pasha (Amirpasha Mozaffari) and co-authors map the risks and propose a fairer path: arxiv.org/abs/2603.05710 #ClimateAI
Exascale weather and climate models produce vast data volumes. The challenge is reducing them without losing the signals science needs. ESiWACE3’s Data Compression Hackathon brings modellers and compression experts together. Online: events.ecmwf.int/event/526/
Working on HPC, AI or large-scale methods for environmental & Earth sciences? #HPC4EES at SC26 is an opportunity to share your work, connect across communities and help shape the field. Pleased to be part of the organising team. Submit by 31 July: hpc4ees.github.io/