Tópicos Especiais em Micropaleontologia/Special Topics in Micropaleontology
In recent years, recognizing changes in the oceans, occurring on different time scales and resulting from natural and anthropogenic events, has become increasingly important. This discipline, therefore, contributes to the use of tools applicable to understanding the reconstruction of temperature, productivity, and oxygenation, among other parameters, in the oceans. The discipline also drives innovation through new technologies in microfaunal analysis, enabling the identification of variations in these environmental parameters. This intensive course follows a single thread, from the living organism to the environmental parameter in a paleoclimate record, using planktonic foraminifera as the model group. Over three class mornings students build the conceptual foundation (biology and ecology; proxies and paleoceanographic reconstruction; imaging and machine learning). Four full practical days then put it into practice: a hands-on day on the Sashimi automated imaging system and the ParticleTrieur classification pipeline, and a field day using some devices for water and sediment sampling. Participants leave able to acquire, image, classify, and ecologically interpret a foraminiferal dataset with a reproducible, semi-automated workflow. The course will be held in person and in English on a full-time basis from September 22 to 25, 2026, at the USP Oceanographic Institute, with fieldwork or practical classes taking place from September 28 to 30.
Organizador(es): Silvia Helena de Mello e Sousa , Sonia Chaabane , Thibault de Garidel-Thoron
Patrocinadores:
- International Research Center USP-CNRS
- Instituto de Oceanografia (IO), Universidade de São Paulo (USP)
Sessões
-
Modern planktonic foraminifera - biology & ecology
3h class What we do: • Foraminifera in the tree of life; the plankton-to-proxy-to-prediction pipeline (course roadmap). • Cell biology, chamber formation, calcification; spinose vs. non-spinose forms; symbiosis. • Life cycle, gametogenesis, lunar/seasonal reproduction; depth habitat and migration. • Ecology & biogeography: hydrographic provinces, seasonality, flux; living vs. dead distinction. • Short scope session: reference-slide morphology under the stereomicroscope. By the end of the day you can: • Recognize major morphogroups and place them phylogenetically. • Relate assemblage composition to oceanographic conditions.
-
Proxies & paleoceanographic reconstruction
What we do: • Stable isotopes: δ18O (temperature + ice volume) and δ13C (productivity/ventilation). • Mg/Ca paleothermometry; cleaning, dissolution, and calibration; vital effects. • Assemblage transfer functions, morphometrics, and coiling ratios. • Multi-proxy reconstruction and error; brief case studies (glacial–interglacial, deglaciation). • Guided problem session: compute a simple SST record from supplied δ18O / Mg/Ca data. By the end of the day you can: • Explain the main proxies and their assumptions. • Sketch a reconstruction from proxy data with realistic uncertainties.
-
Imaging techniques & machine-learning workflows
· Imaging techniques & machine-learning workflows - 3 h class What we do: • Imaging modalities: stereomicroscopy, SEM, focus stacking, hyperfocal (z-stack) scanning. • Building a clean image dataset: annotation, class balance, augmentation, data leakage. • CNN fundamentals; object detection (Faster R-CNN) vs. classification (ResNet); transfer learning. • Train/validation/test splits; metrics - accuracy, precision, recall, F1; confusion matrices; the 'unsure' class. • Optional short Python/notebook primer (no prior coding needed) to prepare for Day 4. By the end of the day you can: • Choose an appropriate imaging method and dataset design. • Read training curves and confusion matrices critically.
-
Fieldwork preparation
What we do: • Define the scientific objectives of a field campaign. • Select suitable sampling stations using maps, bathymetry, and available environmental data. • Prepare the sampling plan, station sheets, and metadata templates. • Check and prepare the field equipment and consumables. • Label the sample containers and organize sample storage. • Assign team roles and organize transport, safety, and field logistics. By the end of the session you can: • Design a coherent sampling plan adapted to the scientific objectives. • Prepare the equipment, documentation, and logistics required for field sampling.
-
Hands-on - Sashimi imaging & ParticleTrieur
What we do: • AM - Sashimi: hardware tour (modified 3D-printer gantry, telecentric optics, ring light); automated slide scanning; z-stacking and image fusion. • AM - object detection to crop single-specimen images from fields of view. • PM - ParticleTrieur: labeling, K-NN mislabel detection, training a ResNet classifier. • PM - running inference at a confidence threshold; handling 'unsure' specimens; exporting counts. • Sanity check: compare automated counts against a manual identification of the same slide. By the end of the day you can: • Scan a slide and produce fused, single-specimen images on Sashimi. • Train and run a classifier in ParticleTrieur and export usable counts.
Enviado pela conta: fmlt