- Prepare seismic features based on our existing methods
- Validate trained models with different types of seismic features
- Statistical analysis of model performance for different catchments
- Presentation of results in our group seminar
- Enrolment at a German University
- B.Sc./ M.Sc. in Science, Engineering, or Computer Science
- Basic knowledge of data science and time series data analysis
- Basic knowledge of Linux, Slurm, Pytorch, Scikit-learn
- Fluent in English
- State-of-the-art equipment
- Flexible working hours and conditions
- Working at the Albert Einstein Science Park on the Telegrafenberg in Potsdam
- Ambitious and varied tasks in a dynamic and international research environment
- Workplace within walking distance of Potsdam main train station or just a short ride on the shuttle bus Start date: As soon as possible
Student Assistant - Potsdam, Deutschland - Helmholtz Centre Potsdam
Beschreibung
Reference Number
9303Section 4.7 "Earth Surface Process Modelling" at the GFZ focuses on developing models to simulate the processes responsible for the evolution of the Earth's surface. In the Hazards and Surface Processes Research Group, we aim to understand the mechanism of different kinds of natural hazards, such as landslides, rockfalls, floods, and debris flows, and migrate these hazards
"Testing and optimising machine-learning-driven debris flow early warning systems''
Debris flow is a widespread gravity-driven hazard in mountainous regions, mixed with water, mud, and solid particles of various sizes. Due to the high impact force, strong basal-lateral erosion, and long outflow distances, debris flow emerged as one of the most destructive mass movements in the mountains and posed a huge threat to residents and property along the main flow path. Early warning systems offer a cost-effective approach to debris flow risk mitigation. Furthermore, environmental seismology, which focuses on recording and interpreting the mechanical vibrations at the Earth's surface, provides a novel, non-invasive, high-temporal resolution method for mass movement monitoring. Our previous work has shown that machine learning can improve the accuracy of identifying debris flow in seismic signals and extend early warning times. Machine learning models were trained to detect debris flow from seismic signals, and a warning strategy was designed for the Illgraben catchment in Switzerland. To test the migration ability of the models, new seismic data from different catchments will be used.
Section 4.7 supports increasing the number of female researchers and, therefore, particularly invites interested women to apply. The student assistant will be working with Dr. Hui Tang and Qi Zhou.
Your responsibilities:
Your qualifications:
What we offer:
Fixed-term: 3 months, with an option to extend (this requires a valid enrollment certificate)
Salary: According to the rules of the TdL (tariff area East) for student assistants, currently € without a bachelor's degree) and € with a bachelor's degree)
Working hours: 10-20 hours per week
Place of work: Potsdam
Are you interested?
If so, we look forward to receiving your application by 28th April 2024 . Please use our online application form only.