Somogyvári et al. (2025) Regional-scale groundwater analysis with dimensionality reduction

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2025
• Date: 2025-11-24
• Authors: Márk Somogyvári, Fabio Brill, Mikhail Tsypin, Lisa Rihm, Tobias Krueger
• DOI: 10.5194/nhess-25-4613-2025

Research Groups

• Integrative Research Institute on Transformations of Human-Environment Systems (IRI THESys), Humboldt-Universität zu Berlin, Berlin, Germany
• Geography Department, Humboldt-Universität zu Berlin, Berlin, Germany
• GFZ German Research Centre for Geosciences, Geosystems Department, Section 4.5 Subsurface Process Modeling, Potsdam, Germany
• Institute of Applied Geosciences, Technische Universität Berlin, Berlin, Germany

Short Summary

This paper proposes a novel methodology using dimensionality reduction on the misfits between modeled and observed groundwater levels to analyze regional-scale climate effects on groundwater changes. The approach successfully identifies regions with distinct climate-groundwater relations and vulnerabilities in the Berlin-Brandenburg area, demonstrating that linear models can capture monthly groundwater dynamics even under significant anthropogenic influence.

Objective

• To delineate regions with different climate-groundwater interactions by applying dimensionality reduction to the misfits between linear water balance model simulations and observed groundwater levels.
• Hypothesis: In temperate climates with periglacial geomorphology, groundwater response to weather forcings can be modeled linearly, with local anomalies in misfits indicating strong anthropogenic influence or unique environmental/geological conditions.

Study Configuration

• Spatial Scale: Regional scale, covering the Berlin-Brandenburg region, Germany, plus a 30 km buffer zone, analyzed on a 2 km × 2 km grid.
• Temporal Scale: Monthly resolution, covering the period from 1990 to 2023.

Methodology and Data

• Models used:
  • Linear water balance models (independent for each grid cell)
  • Dimensionality reduction techniques: Principal Component Analysis (PCA) and Multidimensional Scaling (MDS)
  • Universal Kriging (for spatial interpolation of groundwater levels)
  • Gaussian mixture model (for clustering dimensionality reduction results)
• Data sources:
  • Gridded monthly precipitation and actual evapotranspiration data from the CER v2 dataset (2 km × 2 km spatial grid).
  • Monthly groundwater level data from 504 monitoring wells in Berlin (Berliner Senatsverwaltung für Umwelt, Mobilität, Verbraucher- und Klimaschutz) and Brandenburg (Ministerium für Landwirtschaft, Umwelt und Klimaschutz), interpolated to a 2 km × 2 km grid using Universal Kriging.

Main Results

• Climate-groundwater relations in the Berlin-Brandenburg region are predominantly linear at a monthly scale, even in areas with strong anthropogenic influences.
• The proposed methodology effectively regionalizes groundwater behavior by analyzing model misfits, highlighting areas with anomalous dynamics.
• The first three principal components of the misfit time series explain 85% of the variance (42%, 25%, and 8% respectively).
• The first principal component primarily highlights anthropogenic influences, such as the former open-pit mining region of Lausitz and urban areas like Berlin.
• The second principal component shows correlation with regional topography, distinguishing lowlands (discharge areas) from highlands (recharge areas).
• A systematic overestimation of groundwater levels by the linear models was observed during the last 10 years (a period of decreasing trends), suggesting a change in groundwater response to climatic drivers.
• The method successfully identifies areas with different groundwater dynamics, including river alluvial plains (good model fit), urban areas (worse fit due to altered recharge), post-mining areas (significant anthropogenic level increase still captured by linear model dynamics), and elevated areas with thick vadose zones (poor fit for multiyear periodicity).

Contributions

• Introduces a novel data-driven framework for regional-scale groundwater assessment by applying dimensionality reduction to the misfits between simple water balance models and observed groundwater levels, rather than directly to data or model parameters.
• Provides a method to identify and delineate regions with distinct climate-groundwater interactions and anomalous behavior without requiring detailed hydrogeological knowledge or relying on natural environmental boundaries.
• Demonstrates the utility of simple linear models for capturing groundwater dynamics at a regional scale in temperate, peri-glacial environments, serving as an initial step in a downward model development approach.
• Offers a robust tool for initial investigations in data-rich but knowledge-scarce regions, facilitating more focused and complex modeling studies.

Funding

• Einstein Research Unit “Climate and Water under Change” from the Einstein Foundation Berlin and Berlin University Alliance (grant no. ERU-2020-609).
• Open Access Publication Fund of Humboldt-Universität zu Berlin.

Citation

@article{Somogyvári2025Regionalscale,
  author = {Somogyvári, Márk and Brill, Fabio and Tsypin, Mikhail and Rihm, Lisa and Krueger, Tobias},
  title = {Regional-scale groundwater analysis with dimensionality reduction},
  journal = {Natural hazards and earth system sciences},
  year = {2025},
  doi = {10.5194/nhess-25-4613-2025},
  url = {https://doi.org/10.5194/nhess-25-4613-2025}
}