Robinson et al. (2025) The hydrological archetypes of wetlands

Identification

• Journal: Hydrology and earth system sciences
• Year: 2025
• Date: 2025-11-04
• Authors: Abigail Robinson, Anna Scaini, Francisco J. Peña, Peter A. Hambäck, Christoph Humborg, Fernando Jaramillo
• DOI: 10.5194/hess-29-5975-2025

Research Groups

• Department of Physical Geography, Stockholm University, Stockholm, Sweden (Abigail E. Robinson, Anna Scaini, Fernando Jaramillo)
• Division of Health Informatics and Logistics, KTH Royal Institute of Technology, Stockholm, Sweden (Francisco J. Peña)
• Department of Ecology, Environment and Plant Sciences, Stockholm University, Stockholm, Sweden (Peter A. Hambäck)
• Baltic Sea Centre, Stockholm University, Stockholm, Sweden (Christoph Humborg)

Short Summary

This study investigated the hydrological regimes of 43 Ramsar wetlands in Sweden using Sentinel-1 SAR imagery and a deep learning model (DeepAqua) to predict surface water extent. It identified five distinct hydrological archetypes, revealing how wetland water dynamics relate to their ecosystem services and providing a novel classification method.

Objective

• To investigate and categorize the hydrological regimes of wetlands based on water surface extent observations to better understand their hydrological functions and associated ecosystem services.

Study Configuration

• Spatial Scale: 43 Ramsar sites across Sweden, with individual site areas ranging from 200 hectares to 28,900 hectares.
• Temporal Scale: January 2020 to August 2023, with hydrological regime analysis focused on monthly averages from March to October.

Methodology and Data

• Models used:
  • DeepAqua: A self-supervised deep learning image segmentation model for surface water detection from Sentinel-1 SAR imagery.
  • K-means cluster analysis: Used to group wetlands based on hydrological similarity, employing parameters such as skewness, kurtosis, normalized maximum slope, number of peaks, and baseline month fraction.
• Data sources:
  • Remote Sensing: Sentinel-1 SAR imagery (VH polarization, C-band) for water extent detection; Sentinel-2 optical imagery for Normalized Difference Water Index (NDWI) binary masks (used for DeepAqua training).
  • Environmental Data: Digital Elevation Model 50 m (Markhöjdmodell Nedladdning, grid 50+) from Lantmateriet for elevation; Copernicus Climate Change Service E-OBS ensemble (0.1° grid) and Precipitation Temperature Hydrological Agency's Water Model (PTHBV) from SMHI for precipitation and temperature data.
  • Wetland Classification: Ramsar Convention database, National Wetland Inventory for Sweden (VMI), and an updated satellite-based open wetland mapping classification (Hahn and Wester, 2023).
  • Validation Data: Manually delineated water extent from Sentinel-1 SAR imagery; daily discharge data from Global Runoff Data Centre (GRDC) and SMHI.

Main Results

• DeepAqua's predicted water extent showed strong agreement with manually delineated water extent (Normalized Root Mean Square Error (NRMSE) between 0.04 and 0.12) and generally replicated the shape of hydrological regimes when compared to in situ daily discharge data.
• Five distinct hydrological archetypes were identified for the 43 Ramsar sites:
  1. Spring-surging (n=6): Flashy regimes with a dry baseline and a brief period of increased water extent in spring, predominantly found in northern Sweden.
  2. Spring-flooded (n=8): Similar to spring-surging but with a relatively longer spring peak, located in southern and central Sweden.
  3. Summer-flooded (n=8): Wetlands remaining inundated from May to October after a rapid wetting period, distributed across Sweden.
  4. Slow-drying (n=15): Exhibiting steadily decreasing water extent throughout summer, reaching minimum in autumn, typical of southern Sweden.
  5. Summer-dry (n=6): Maximum water extent in April, followed by generally dry conditions until September–October.
• Flashy archetypes (e.g., spring-surging, summer-flooded) with high water extent variability were preferentially found at higher elevations and latitudes, while smoother, less variable, and drier archetypes (e.g., slow-drying, summer-dry) were concentrated in central and southern Sweden at lower elevations.
• The archetypes revealed remarkable similarities in the timing and duration of flooding and drying events among grouped wetlands, even across diverse wetland types.
• Hydrological regimes can indicate ecosystem services: spring-surging wetlands (resembling headwater wetlands) may accentuate floods and droughts, while slow-drying wetlands (typical of floodplain wetlands) are more likely to provide flood attenuation and water storage during low flow conditions.

Contributions

• Novel classification of wetlands into hydrological archetypes based on remotely sensed water surface extent, offering a new perspective on wetland hydrological functioning.
• Application and validation of a self-supervised deep learning model (DeepAqua) for robust surface water detection in wetlands using SAR imagery, addressing data scarcity challenges.
• Demonstration that grouping wetlands by hydrological regime provides more comprehensive insights into their functions and associated ecosystem services than traditional wetland type classifications.
• Proposal of a methodology for hydrological-regime-based wetland classification that is easily applicable to other wetland-rich landscapes, facilitating better understanding and management of ecosystem services.
• Elucidation of the spatial distribution of different hydrological archetypes across Sweden and their links to environmental characteristics (e.g., latitude, elevation).

Funding

• Svenska Forskningsrådet Formas (grant no. 2022-01570)
• Project DOWES “Disclosing the Overlooked Wetlandscape Ecosystem Services” of the Water4All 2023 Joint Transnational Call on Aquatic Ecosystem Services, handled by FORMAS (project number 2024-00999)
• Swedish Research Council, Forte, Formas, and Vinnova (for article publication)

Citation

@article{Robinson2025hydrological,
  author = {Robinson, Abigail and Scaini, Anna and Peña, Francisco J. and Hambäck, Peter A. and Humborg, Christoph and Jaramillo, Fernando},
  title = {The hydrological archetypes of wetlands},
  journal = {Hydrology and earth system sciences},
  year = {2025},
  doi = {10.5194/hess-29-5975-2025},
  url = {https://doi.org/10.5194/hess-29-5975-2025}
}