Liu et al. (2026) Comparative performance of random forest and support vector regression for reference evapotranspiration estimation: a critical review

Identification

• Journal: Frontiers in Environmental Science
• Year: 2026
• Date: 2026-09-08
• Authors: Chenchen Liu, Nor Azimah Khalid, Muhammad Izzad Ramli
• DOI: 10.3389/fenvs.2026.1895651

Research Groups

• Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
• School of Information and Technology, Xichang College, Xichang, Sichuan, China

Short Summary

This review compares the performance of random forest (RF) and support vector regression (SVR) models for reference evapotranspiration estimation under data-scarce conditions. The study synthesizes 48 peer-reviewed primary studies to evaluate the predictive capabilities of both algorithms across diverse climatic conditions and varying data availability.

Objective

• Investigate the comparative performance of RF/RFR and SVM/SVR models for ET0 and evapotranspiration-related prediction tasks.
• Evaluate the limitations of these machine learning models under data-scarce conditions.

Study Configuration

• Spatial Scale: Global, with a focus on diverse climatic conditions.
• Temporal Scale: 2019 to 2026, covering recent literature on precision irrigation.

Methodology and Data

• Models used:
  • Support Vector Regression (SVR)
  • Random Forest (RF)
• Data sources:
  • Peer-reviewed journal articles published in English
  • Web of Science, Scopus, and IEEE Xplore databases

Main Results

• Comparative evaluations suggest that apparent model superiority is conditional on data dimensionality, predictor availability, target construction, and validation strategy.
• SVR may be competitive in limited-input, low-dimensional, or relatively stable settings, whereas RF/RFR may be advantageous for heterogeneous, multi-source, or spatially distributed predictor sets.

Contributions

• This review provides a critical evaluation of the comparative performance and limitations of SVM/SVR and RF/RFR models for ET0 and evapotranspiration-related prediction tasks.
• The study highlights the importance of considering data dimensionality, predictor availability, target construction, and validation strategy when selecting machine learning models for precision irrigation.

Funding

• This research was supported by the Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia.

Citation

@article{Liu2026Comparative,
  author = {Liu, Chenchen and Khalid, Nor Azimah and Ramli, Muhammad Izzad},
  title = {Comparative performance of random forest and support vector regression for reference evapotranspiration estimation: a critical review},
  journal = {Frontiers in Environmental Science},
  year = {2026},
  doi = {10.3389/fenvs.2026.1895651},
  url = {https://doi.org/10.3389/fenvs.2026.1895651}
}