Integrating Deep Learning and Data Assimilation for High-Resolution Turbidity Modeling in
the European Digital Twin Ocean.
A consortium of research institutes is building the underlying models for the European Digital Twin Ocean. Multiple
hydrodynamic model framworks will be improved and integrated into a virtual ocean model. To enhance the physics-
based models, Deep Differentiable Emulators (AI components) will be implemented in the model framework. The
project is aligned with the EU mission: Restore our ocean, which aims to achieve zero pollution, net zero emission
and biodiversity restoration.
In collaboration with a French project partner, IMT Atlantique, you will apply 4DVarNET (Fablet et al., 2021) to
construct 2D/3D turbidity fields at high resolution (~100 m). The demonstration site is the Dutch Wadden Sea, a
marine protected area and World Heritage site. You will explore task-optimised trainable variational data assimilation
models and solvers, with uncertainty quantification for partial ocean states, typically sea surface tracers. You will
implement a demonstration in observation-driven forecasting and reanalysis, with a focus on biogeochemical
tracers, coastal turbidity and sea surface dynamics. The general objective is to bridge model-based data assimilation
and learning paradigms, with a focus on uncertainty quantification.
Enrolled in a relevant MSc program software development, data science, earth observation,
or innovative technologies for modeling or monitoring.
To facilitate collaboration with Dutch universities for our internship program, we encourage students who are currently enrolled in a Dutch university and residing in the Netherlands to apply.
Program duration is 6 – 9 months starting September 2023.
Interning with Deltares means a monthly student allowance of €550 (depending on your
study points), flexible working hours, and the chance to work alongside top researchers from a Triple A institute.
Don’t miss this unique opportunity to make a difference!Solliciteren