Jobbeschreibung
The Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research (AWI) is a member of the Helmholtz Association (HGF) and funded by federal and state government. AWI focuses on polar and marine research in a variety of disciplines such as biology, oceanography, geology, geochemistry and geophysics thus allowing multidisciplinary approaches to scientific goals.
Postdoc “Remote Sensing and Geospatial Data Science for Soil Carbon Representation in the 3D-ABC Foundation Model” (m/f/d)
Background
The large collaborative project ‘3D-ABC', funded for 3 years by the Helmholtz Foundation Model Initiative, will focus on the development of a Foundation Model combining large remote sensing and field datasets, large-scale AI, generative AI, and Exascale HPC to detect, quantify, and characterize key parameters of the global carbon cycle at high spatial resolution with a focus on above and below ground terrestrial carbon stocks.
3D-ABC is led by AWI and combines the resources and expertise from 6 HGF centres (AWI, GFZ, UFZ, DLR, FZJ, and HZDR). It will build strongly on HGF-associated datasets including products from the TanDEM-X radar mission, large airborne and field-based LiDAR datasets, and process-based models, as well as other openly available remote sensing and field datasets. The diverse data will be combined to facilitate 8 downstream tasks, from land cover change and forest height, to increasingly demanding tasks such as above-ground biomass, soil carbon storage, and model uncertainty ranges, to eventually complex tasks targeting carbon impacts of disturbances, forest productivity, and carbon uptake. Training the 3D-ABC Foundation Model involves extremely large volumes of data and computational resources, requiring High Performance Computing (HPC) on the JUWELS HPC system and the upcoming first European Exascale supercomputer, JUPITER (both at FZJ) to facilitate exceptionally high performance for AI training and deployment at the scale needed for 3D-ABC.
The postdoc position requires experience in handling and processing of large remote sensing, various geospatial, and field datasets with a focus on global soil carbon stocks. You will focus on preparing these datasets for ingestion into the 3D-ABC Foundation Model, analyse geospatial soil carbon covariates such as remotely sensed land cover, geomorphology, climate and multispectral indices, and support the development of the below-ground carbon storage downstream task of the 3D-ABC Foundation Model.
You will lead the development of soil covariate datasets for the Foundation Model, design aspects of the global soil dataset analysis, and manage an efficient and suitable transfer of the data patches for Foundation Model training to the AI and HPC teams in the 3D-ABC consortium.
- Lead the collation and aggregation of global soil carbon datasets for the 3D-ABC Foundation Model database
- Lead the development and analysis of soil carbon covariates from remote sensing and other geospatial datasets
- Develop a database of soil carbon data patches for training the 3D-ABC Foundation Model in close collaboration with the project AI and HPC team
- Publish on the soil carbon synthesis and remote sensing analysis datasets
- You will be embedded in the Permafrost Research Section in Potsdam and will be able to use the rich data,
- Hardware and software resources available at AWI. Additional computing resources are available at the FZJ HPC centre.
- Completed doctoral degree in the field of geospatial sciences, remote sensing, geoinformatics, soil sciences, physical geography, or other closely related fields
- Demonstrated experience in handling, processing, and analysing large remote sensing image and other geospatial datasets
- Experience in programming and in automation of data processing (image data, maps, soil data)
- Interest in the development and analysis of soil carbon synthesis datasets, including for large-scale AI and FM training
- First experience with using GPU, HPC, and/or AI processing methods and dataset requirements
- Relevant publication record and experience presenting scientific results at international level
- Ability and willingness to publish research results in the peer-review literature
- Ability and willingness to work focused, independently, and cooperatively in a large team and with international partners
- Excellent communication skills and fluent knowledge of English (spoken and written)
- Experience in the presentation of results at international conferences
Additional skills and knowledge
- Experience in Deep Learning and Foundation Models, e.g. Prithvi, MajorTom, or BigEarthNet, etc.
- Prior experience with permafrost regions and permafrost soil carbon datasets and their unique characteristics
- German language knowledge
- our scientific success - excellent research
- collaboration and cooperation - intra-institute, national and international, interdisciplinary
- opportunities to develop – on the job and towards other positions
- an international environment – everyday contact with people from all over the world
- flexible working hours and the possibility of mobile working up to 50% of regular working hours
- health promotion and company fitness
- support services and a culture of reconciling work and family
- occupational pension provision (VBL)