Articles | Volume 18, issue 3
https://doi.org/10.5194/essd-18-1783-2026
https://doi.org/10.5194/essd-18-1783-2026
Data description article
 | 
10 Mar 2026
Data description article |  | 10 Mar 2026

Reconstructing nineteenth-century Danube river water levels with transformer-based computer vision

Malte Rehbein

Data sets

Bavarian Danube Water Level Reconstruction (1826-1894) Malte Rehbein https://doi.org/10.5281/zenodo.17296750

Model code and software

LineFormer: Line Chart Data Extraction Using Instance Segmentation (https://github.com/TheJaeLal/LineFormer) J. Lal et al. https://doi.org/10.1007/978-3-031-41734-4_24

Download
Short summary
We transformed nineteenth-century hand-drawn Bavarian Danube gauge charts into daily water-level records using a largely automated, human-checked image analysis workflow. Tests at several gauges show the method reproduces levels with high accuracy while significantly cutting manual effort. The resulting open scientific datasets and traceable sources enable researchers and planners to study past floods and droughts, improve long records, and inform river management and climate impact assessments.
Share
Altmetrics
Final-revised paper
Preprint