Oh et al. (2025) Scalable, adaptive and risk-informed design of hydrological sensor networks

Identification

• Journal: Nature Water
• Year: 2025
• Date: 2025-10-07
• Authors: Jeil Oh, Matthew Bartos
• DOI: 10.1038/s44221-025-00496-7

Research Groups

Fariborz Maseeh Department of Civil, Architectural and Environmental Engineering, The University of Texas at Austin, Austin, TX, USA

Short Summary

This study introduces a data-driven framework for designing streamflow monitoring networks that enhances hydrological predictions and integrates socio-environmental constraints. The framework, utilizing rank-revealing QR decomposition, demonstrates superior streamflow reconstruction at ungauged locations compared to existing methods, while also being scalable and adaptable to flood risk.

Objective

• To develop a scalable, adaptive, and risk-informed data-driven framework for designing streamflow monitoring networks that improves hydrological predictions and accommodates socio-environmental constraints.

Study Configuration

• Spatial Scale: Diverse hydrological regimes; continental-scale (implied by National Water Model and various global/regional datasets).
• Temporal Scale: 44 years of retrospective reanalysis data.

Methodology and Data

• Models used: Rank-revealing QR decomposition for isolating optimal monitoring sites.
• Data sources:
  • National Water Model Retrospective Dataset (reanalysis)
  • GloFAS (Global Flood Awareness System)
  • USGS gauge inventory (observation)
  • CAMELS-BR (hydrometeorological time series)
  • Bangladesh Water Development Board (discharge data)
  • USGS National Hydrography Dataset Plus v2.1 (geographic)
  • FEMA National Risk Index (geographic, risk data)
  • PRISM climate data (climate data)
  • SEDAC population density data (socio-environmental data)

Main Results

• The proposed sensor placement approach enables better reconstructions of streamflow at ungauged locations compared with existing methods.
• The framework effectively accommodates incremental expansion of existing gauge networks.
• It successfully integrates operational priorities, such as flood risk, without compromising the accuracy of hydrological predictions.
• The framework is demonstrated to be scalable and robust across diverse hydrological regimes.

Contributions

• Presents a novel data-driven framework for streamflow monitoring network design using rank-revealing QR decomposition.
• Offers a method that significantly improves hydrological predictions at ungauged locations compared to current approaches.
• Introduces adaptability for incremental network expansion and integration of socio-environmental constraints like flood risk into network design.
• Provides a scalable and robust tool for water managers to make more informed decisions regarding sensor network deployment.

Funding

• University Graduate Continuing Fellowship from the University of Texas at Austin (supported J.O.).

Citation

@article{Oh2025Scalable,
  author = {Oh, Jeil and Bartos, Matthew},
  title = {Scalable, adaptive and risk-informed design of hydrological sensor networks},
  journal = {Nature Water},
  year = {2025},
  doi = {10.1038/s44221-025-00496-7},
  url = {https://doi.org/10.1038/s44221-025-00496-7}
}