E+ Blended Intensive Programme: “Artificial Intelligence for Forest and Landscape Remote Sensing”
Course description (ENLIGHT)
Environmental
Data Science
Data Science
What we do
We connect methodological developments in data science with application-oriented research questions from the environmental, ecosystem, and forest sciences.
About the Environmental Data Science Section
Data-driven methods and AI applications are advancing rapidly in the environmental, ecosystem, and forest sciences and are becoming a central component of research. At the same time, solid data science and AI skills are gaining importance for students in these fields — for academic careers as well as for entering the job market outside academia.
“A bridge between data-driven methodological foundations and concrete environmental and forestry applications.”
The particular added value of the section lies in the systematic combination of methodological developments in data science with application-oriented research questions from the forest and environmental sciences. In doing so, it serves both as a network and as a thematic platform for application-oriented research.
What are our goals
- Coordinating and connecting research activities in data-driven environmental and forest research.
- Promoting cross-faculty collaborations between application and methods disciplines.
- Supporting joint third-party funding applications, particularly in interdisciplinary collaborative projects.
- A framework for joint theses and doctoral projects across faculty boundaries.
- A point of contact for doctoral researchers at this interface — for exchange, co-supervision, and colloquia.
- Workshops and seminars.
- Developing cross-faculty elective modules and teaching formats in the field of environmental data science.
Courses
Theses
Master's thesis
From Prompts to Reproducible Forest Inventory Analysis: LLM-Orchestrated Forest Inventory Analysis
Background: Forest inventory analysis increasingly combines tabular data, established domain methods, statistical workflows and specialist software. Large language models (LLMs) offer a new way to access such analyses through natural language. However, direct LLM-based calculations may be unreliable and difficult to reproduce. A promising alternative is to let an LLM translate user prompts into calls to validated, deterministic analysis tools.
Objectives: This thesis will develop and evaluate an AI-assisted system that translates natural-language requests into reproducible forest inventory analyses. Combining forestry expertise with modern data science, the system will use an LLM to select and orchestrate validated Python/R tools for tasks such as data validation, stand metrics, structural analysis, species composition and volume estimation.
Prerequisites: Knowledge of forest inventory and basic statistics; willingness to self-study Python; interest in AI/LLMs and reproducible data analysis.
Thesis description (PDF)What is being planned?
Seminar: Use of Sentinel-2 Super-resolution imagery
Seminar: Obtain information from satellite Embeddings
Do you conduct research at this interface?
The Environmental Data Science Section welcomes new collaborations from the environmental, ecosystem, and forest sciences as well as related methods disciplines. Please contact the section speaker, Dr. Nils Nölke (Nils.Noelke@forst.uni-goettingen.de)