Ma et al. (2026) Short-, medium-, and long-term reference evapotranspiration forecasting based on intelligent temperature forecast correction and regional calibration of the Hargreaves model

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2026
• Date: 2026-09-23
• Authors: Zhongxin Ma, Xin Han, Baozhong Zhang, Xin Wang, Qian Huang, Zhigong Peng, Kai Zhang, Xin Li
• DOI: 10.1016/j.ejrh.2026.103999

Research Groups

• College of Water Conservancy Engineering, Tianjin Agricultural University
• State Key Laboratory of Water Cycle and Water Security, China Institute of Water Resources and Hydropower Research
• Water Resources Research Institute of Shandong Province
• Tianjin Agricultural University-China Agricultural University Joint Smart Water Conservancy Research Center

Short Summary

This study evaluates the spatiotemporal characteristics of maximum (T′max) and minimum (T′min) temperature forecast errors over 1–40-day lead times using data from 2019 to 2023. Four temperature forecast correction methods were compared, and the optimal correction method was selected for each lead time.

Objective

• Evaluate the spatiotemporal characteristics of T′max and T′min forecast errors across different lead times.
• Compare four temperature forecast correction methods (ACF-DAM, SR, GA-LSTM, and GA-RF) to select the optimal correction method for each lead time.

Study Configuration

• Spatial Scale: The study area is located in the Yellow River Irrigation Area of Shandong Province, China, with geographical coordinates of 34◦33′–38◦16′N and 114◦48′–119◦19′E.
• Temporal Scale: The study period spans from 2019 to 2023.

Methodology and Data

• Models used:
  • ACF-DAM (Decaying-Average method based on Autocorrelation Function)
  • SR (Stepwise Regression)
  • GA-LSTM (Genetic Algorithm-optimized Long Short-Term Memory network)
  • GA-RF (Genetic Algorithm-optimized Random Forest)
• Data sources:
  • Historical daily meteorological data from 44 stations in the Yellow River Irrigation Area of Shandong Province
  • Forecast data for the same area and period obtained from the China Weather Network

Main Results

• The optimal correction schemes reduced the mean station-wise RMSE of corrected T′max and T′min by 11.2096%–40.0618% and 2.2416%–21.3531%, respectively, relative to the raw forecasts.
• The 3-day lead time yielded the best ETo forecasting performance, with a cross-station mean RMSE of 0.8917 mm d−1.

Contributions

• This study addresses the limited application of lead-time-specific temperature forecast error correction in regional ETo forecasting.
• The results provide insights into the spatiotemporal characteristics of temperature forecast errors and their impact on ETo predictions.

Funding

• This research was funded by [project name], [program name], and [reference code].

Citation

@article{Ma2026Short,
  author = {Ma, Zhongxin and Han, Xin and Zhang, Baozhong and Wang, Xin and Huang, Qian and Peng, Zhigong and Zhang, Kai and Li, Xin},
  title = {Short-, medium-, and long-term reference evapotranspiration forecasting based on intelligent temperature forecast correction and regional calibration of the Hargreaves model},
  journal = {Journal of Hydrology Regional Studies},
  year = {2026},
  doi = {10.1016/j.ejrh.2026.103999},
  url = {https://doi.org/10.1016/j.ejrh.2026.103999}
}