Dr. Josefine Wilms

Josefine Wilms
Dr. Josefine Wilms
Haus 401, Raum 1.05
Claude-Dornier-Str. 1
82234 Weßling

Funktion und Aufgaben:

Applied mathematician performing gravity field retrieval simulations

Wissenschaftliche Interessen:

Gravity field retrieval
Data science and Applied Machine Learning with scarce data
High performance computing


2021-Present Scientist at GFZ
2019-2020 Research Scientist at Deltares, The Netherlands
2015-2018 Sr. Research Scientist at the Council for Scientific and Industrial Research, South Africa.
2013-2015 Post doctoral researcher at Mechatronical Engineering, University of Stellenbosch, South Africa
2008-2012 University of Stellenbosch, South Africa: Part time lecturer for Dynamics second year course and Computational Fluid Dynamics post-graduate course, Applied Mathematics


March 2021 - Present

  • Gravity field retrieval simulations with EPOS-OC for NGGM/MAGIC – Science Support Study and ESA Third Party Mission Study.
  • Support for development of Payload Data Ground Segment concept for the NGGM study.

March 2019 – February 2021

  • Development of SNAP pipeline in python for the calculation of land subsidence from SAR Sentinel1 images.

  • Modelled suspended particle matter in the Wadden sea using stacked ensemble machine learning methods.

  • Developed Flask based API for doing pre-processing for chemical and biological input components to the DFlowFM model.

  • Developed and deployed docker containers for Jupyter notebooks on Azure. These notebooks were for online courses at Deltares.

  • Developed a gradient boosting based machine learning model to predict overtopping at coastal structures.

December 2015-February 2019

  • Developed a river routing model to simulate the movement of surface runoff to oceans on a global scale. 

  • Constructed an interface that converts weather data from three different sources into a standard format and combines these into a single database. 

  • Constructed models to predict overtopping at coastal structures with OpenFOAM (IHFoam).

  • Trained feedforward and recurrent neural networks on river discharge data.

Publikationen werden geladen
zurück nach oben zum Hauptinhalt