Parisouj et al. (2025) Hourly streamflow forecasting across diverse climate zones on Oʻahu Island, Hawaiʻi

Identification

• Journal: Journal of Hydrology
• Year: 2025
• Date: 2025-09-11
• Authors: Peiman Parisouj, Changhyun Jun, Sayed M. Bateni
• DOI: 10.1016/j.jhydrol.2025.134225

Research Groups

• Department of Civil, Environmental, and Construction Engineering and Water Resources Research Center, University of Hawaiʻi at Manoa, Honolulu, HI, USA
• School of Civil, Environmental and Architectural Engineering, College of Engineering, Korea University, Seoul, Republic of Korea
• UNESCO-UNISA Africa Chair in Nanoscience and Nanotechnology College of Graduate Studies, University of South Africa, Muckleneuk Ridge, Pretoria 392, South Africa

Short Summary

This study proposes a novel hybrid Honey Badger Algorithm-optimized Multilayer Perceptron (HBA–MLP) model for hourly streamflow forecasting across diverse climate zones on Oʻahu Island, Hawaiʻi, demonstrating exceptional performance in arid and semi-arid zones while identifying challenges in subhumid regions.

Objective

• To develop and evaluate a novel hybrid HBA–MLP model for accurate hourly streamflow forecasting across diverse climate zones (arid, semi-arid, subhumid, humid) on Oʻahu Island, Hawaiʻi, a task not previously undertaken due to hydrological system complexity.

Study Configuration

• Spatial Scale: Oʻahu Island, Hawaiʻi, focusing on four specific streams: Waikele (arid), Waimea (semi-arid), Makua (subhumid), and Kahana (humid).
• Temporal Scale: Hourly streamflow forecasting.

Methodology and Data

• Models used: Honey Badger Algorithm (HBA) to optimize a Multilayer Perceptron (MLP) model (HBA–MLP).
• Data sources: Streamflow time series data for the four selected streams; Aridity index data for classifying Oʻahu's climate zones.

Main Results

• The HBA–MLP model demonstrated exceptional performance in forecasting streamflow across the diverse climate zones.
• Waikele stream (arid zone): Root Mean Square Error (RMSE) of 0.80 m³/s, Kling–Gupta Efficiency (KGE) of 0.95, Nash–Sutcliffe Efficiency (NSE) of 0.92. Mean absolute relative error (ARE) for peak flows was 17.64 %.
• Waimea stream (semi-arid zone): RMSE of 0.60 m³/s, Coefficient of Determination (R²) of 0.93. Mean ARE for peak flows was 14.44 %.
• Kahana stream (humid zone): RMSE of 0.22 m³/s, KGE of 0.85. Mean ARE for peak flows was 28.78 %.
• Makua stream (subhumid zone): RMSE of 0.08 m³/s, KGE of 0.68. Mean ARE for peak flows was 29.38 %. This zone presented more challenges due to Ultisols and rapid subsurface flow.
• The model achieved reliable forecasting of extreme flow events in arid and semi-arid zones, with moderate accuracy in the humid zone, but experienced greater difficulties in the subhumid zone for peak flow prediction.

Contributions

• First comprehensive study to forecast streamflow across O‘ahu’s diverse climate zones, filling a critical gap in regional forecasting research.
• Introduces a novel hybrid HBA–MLP machine learning model for solving complex hydrological problems.
• Provides valuable insights for improving real-time flood forecasting and water resource management in Hawaii.
• Highlights areas for future improvement, particularly in subhumid regions where complex subsurface flow dynamics and soil characteristics affect forecasting performance.

Funding

• Not explicitly mentioned in the provided text.

Citation

@article{Parisouj2025Hourly,
  author = {Parisouj, Peiman and Jun, Changhyun and Bateni, Sayed M.},
  title = {Hourly streamflow forecasting across diverse climate zones on Oʻahu Island, Hawaiʻi},
  journal = {Journal of Hydrology},
  year = {2025},
  doi = {10.1016/j.jhydrol.2025.134225},
  url = {https://doi.org/10.1016/j.jhydrol.2025.134225}
}