Wang et al. (2026) Spectral-enhanced cross-domain network: a novel daily runoff prediction model based on energy amplification and cross-domain multi-scale collaboration

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-07-13
• Authors: R Y Wang, Tao Yang
• DOI: 10.1016/j.jhydrol.2026.136018

Research Groups

• State Key Laboratory of Water Disaster Prevention, Hohai University, Nanjing 210098, China
• Dayu College, Hohai University, Nanjing 210098, China
• College of Hydrology and Water Resources, Hohai University, Nanjing 210098, China
• Yangtze Institute for Conservation and Development, Hohai University, Nanjing 210098, China

Short Summary

This study proposes the Spectral-Enhanced Cross-domain Network (SEC), a novel daily runoff forecasting model that integrates spectral energy modulation with cross-domain multi-scale collaboration to improve hydrological prediction.

Objective

• Investigate the effectiveness of integrating spectral energy modulation and cross-domain multi-scale collaboration in improving daily runoff forecasting models.

Study Configuration

• Spatial Scale: Catchment scale, representing diverse hydro-climatic regions in the United States.
• Temporal Scale: Daily time step for runoff prediction.

Methodology and Data

• Models used: Spectral Energy Modulator (SEM) and Cross-Domain Multi-Scale Collaboration Mechanism (CDMSCM).
• Data sources: Natural catchments across six representative stations, with data obtained from satellite observations and reanalysis products.

Main Results

• The SEC model achieved maximum NSE and KGE values of 0.977 and 0.986, respectively.
• SEC model reduced MAE by 0.6%–84.9% and RMSE by 2.8%–80.1% compared to nine mainstream benchmark models.

Contributions

• This study demonstrates the utility of spectral energy modulation and cross-domain multi-scale collaboration for modeling non-stationary hydrological sequences.
• The proposed framework provides an improved approach for analyzing extreme hydrological events and has potential for assisting in short-term extreme event forecasting.

Funding

• This research was funded by the State Key Laboratory of Water Disaster Prevention (Grant No. SKLWDP2023001).

Citation

@article{Wang2026Spectralenhanced,
  author = {Wang, R Y and Yang, Tao},
  title = {Spectral-enhanced cross-domain network: a novel daily runoff prediction model based on energy amplification and cross-domain multi-scale collaboration},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136018},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136018}
}