Roy et al. (2026) Applying Transfer Learning for Street‐Scale Nuisance Flood Forecasting in Coastal‐Urban Environments

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Water Resources Research
• Year: 2026
• Date: 2026-08-01
• Authors: Binata Roy, Jonathan L. Goodall, Diana McSpadden, Chetan Kumar, Steven Goldenberg, Yidi Wang, M. Schram
• DOI: 10.1029/2025wr041818

Research Groups

• Department of Civil and Environmental Engineering, University of Virginia
• Norfolk City Flood Control Office

Short Summary

This study investigates the use of Transfer Learning (TL) for street-scale nuisance flood forecasting by exploring whether an ML model trained on one set of streets can effectively forecast flooding for another set of streets in the same city. The results show that TL is most effective when target data is limited.

Objective

• Investigate the transferability of Machine Learning models for street-scale nuisance flood forecasting using Transfer Learning

Study Configuration

• Spatial Scale: Street-level, Norfolk City, Virginia
• Temporal Scale: Short-term (flood events)

Methodology and Data

• Models used: Long Short-Term Memory (LSTM) model
• Data sources: Flood data from water depth sensors in Norfolk City

Main Results

• Full-weight retraining of the ML model proved most effective for flood forecasting
• Minimal retraining of only the output layer was insufficient
• The advantage of TL diminished as target data increased beyond 18 flood events

Contributions

• This study provides insights into the use of Transfer Learning for street-scale nuisance flood forecasting, which can assist cities in implementing accurate forecasts for new sensing locations with limited data records.

Funding

• National Science Foundation (NSF) Grant #1835236: "Urban Flood Sensing and Forecasting"
• Norfolk City Flood Control Office Research Fund

Citation

@article{Roy2026Applying,
  author = {Roy, Binata and Goodall, Jonathan L. and McSpadden, Diana and Kumar, Chetan and Goldenberg, Steven and Wang, Yidi and Schram, M.},
  title = {Applying Transfer Learning for Street‐Scale Nuisance Flood Forecasting in Coastal‐Urban Environments},
  journal = {Water Resources Research},
  year = {2026},
  doi = {10.1029/2025wr041818},
  url = {https://doi.org/10.1029/2025wr041818}
}