Postdoctoral Researcher - AI4PEX - Artificial Intelligence and Machine Learning for Enhanced Representation of Processes and Extremes in Earth System Models
Postdoctoral Researcher - AI4PEX - Artificial Intelligence and Machine Learning for Enhanced Representation of Processes and Extremes in Earth System Models
Max Planck Institute for Biogeochemistry
Jena, Germany
Background and position description
The Atmosphere-Biosphere Coupling, Climate and Causality group (ABC3) and the Model Data Integration group (MDI), of the Department of Biogeochemical Integration (BGI), at the Max Planck Institute for Biogeochemistry in Jena offers a project related position open for a Postdoctoral Researcher in the context of a Horizon Europe project coordinated at the MDI: AI4PEX - Artificial Intelligence and Machine Learning for Enhanced Representation of Processes and Extremes in Earth System Models.
AI4PEX aims to improve the representation of processes in land, ocean and the atmosphere underpinning the largest uncertainties in feedbacks simulated in Earth system models (ESM), thereby reducing uncertainty in climate change projections. AI4PEX is based on a multidisciplinary approach, focused on “learning” from observations to accurately describe processes through a fusion of observations with advanced machine learning (ML) and artificial intelligence (AI). Such data and approaches, constrained by the laws of physics, aims to deliver a step change in the accuracy of ESMs. AI4PEX includes close to 20 partners from different European institutions focused on different aspects of Earth system research: Earth system modeling, Earth observation, AI and ML.
The current position focuses on exploring different ML and AI approaches to develop new ESM representations for key processes on land using Earth observation data. In particular, we are looking to improve biosphere-atmosphere interactions in terms of land carbon uptake and ecosystems’ response to climate extremes in the ICON land surface model from short to long time scales, looking at improving the simulation of ecosystem response to water and heat stress, phenological regimes and shifts in response to climatic changes, up to the dynamics of vegetation mortality and carbon turnover. Conceptually, we aim inducing observation-based ecosystem functioning through hybrid modelling approaches that combine different observational products. Downstream steps include the quantification of model uncertainty via simulation approaches and the coupling of model developments in the ICON Earth system model. In this endeavor, we are anticipating collaborations with CNRS-IPSL (LSCE, Paris), Lund University (Lund), Met Office (UK) and University of Leipzig (Germany).
This is an appointment within the ABC3 and the MDI groups at the BGI. We are involved in development of advanced methods and datasets towards a better understanding of Earth system dynamics. In particular, we are actively involved in researching the use and development of machine and deep learning (ML/DL) approaches to model, parameterize and analyze large datasets on Earth system dynamics, with an emphasis on modeling terrestrial ecosystems and biogeochemical processes. The position relates to the research needs in the context of the AI4PEX project where we bring ML/DL approaches to the core of modeling and analyzing land surface dynamics at the core of the ICON Earth system model.
Your tasks
- Development of a hybrid modeling approach (HM) bridging machine learning and Earth system modeling (ESM)
- Design and perform simulations for uncertainty quantification (perturbed parameter experiments) and in support of project internal requirements and outputs
- Contrast HM to traditional ESM approaches
- Dissemination of scientific results via conference communications and publications in scientific journals
- Assume responsibilities on project reporting and deliverables
Your profile
- Successfully completed PhD in climate science, meteorology, computer science, environmental science, geoecology, biology, geography or comparable fields
- Experience in Earth system modeling and/or land surface process modeling is highly valuable
- Practical experience in scientific programming and data processing (data analysis and processing with e.g. Python, Julia, R; experience in the use of HPC clusters)
- Expertise in statistical analysis, machine learning and/or data assimilation
- Ability to work both independently and as part of a team
- Very good command of written and spoken English
© EuroJobsites 2024
