Open Access

ARTICLE

Forecasting Multi-Step Ahead Monthly Reference Evapotranspiration Using Hybrid Extreme Gradient Boosting with Grey Wolf Optimization Algorithm

by Xianghui Lu, Junliang Fan, Lifeng Wu, Jianhua Dong

1 School of Hydraulic and Ecological Engineering, Nanchang Institute of Technology, Nanchang, 330099, China
2 Key Laboratory of Agricultural Soil and Water Engineering in Arid and Semiarid Areas of Ministry of Education, Northwest A&F University, Yangling, 712100, China
3 Faculty of Agriculture and Food, Kunming University of Science and Technology, Kunming, 650500, China

* Corresponding Author: Lifeng Wu. Email:

Computer Modeling in Engineering & Sciences 2020, 125(2), 699-723. https://doi.org/10.32604/cmes.2020.011004

Received 14 April 2020; Accepted 24 July 2020; Issue published 12 October 2020

Abstract

It is important for regional water resources management to know the agricultural water consumption information several months in advance. Forecasting reference evapotranspiration (ET₀) in the next few months is important for irrigation and reservoir management. Studies on forecasting of multiple-month ahead ET₀ using machine learning models have not been reported yet. Besides, machine learning models such as the XGBoost model has multiple parameters that need to be tuned, and traditional methods can get stuck in a regional optimal solution and fail to obtain a global optimal solution. This study investigated the performance of the hybrid extreme gradient boosting (XGBoost) model coupled with the Grey Wolf Optimizer (GWO) algorithm for forecasting multi-step ahead ET₀ (1–3 months ahead), compared with three conventional machine learning models, i.e., standalone XGBoost, multi-layer perceptron (MLP) and M5 model tree (M5) models in the subtropical zone of China. The results showed that the GWO-XGB model generally performed better than the other three machine learning models in forecasting 1–3 months ahead ET₀, followed by the XGB, M5 and MLP models with very small differences among the three models. The GWO-XGB model performed best in autumn, while the MLP model performed slightly better than the other three models in summer. It is thus suggested to apply the MLP model for ET₀ forecasting in summer but use the GWO-XGB model in other seasons.

---

Keywords

---

Citations