An improved model based on the support vector machine and cuckoo algorithm for simulating reference...

Area of Science:

• Agricultural Science
• Environmental Science
• Data Science

Background:

• Reference evapotranspiration (ET0) is critical for effective irrigated agriculture management.
• Accurate ET0 estimation is essential for optimizing water resource allocation and crop yield.

Purpose of the Study:

• To develop and evaluate an improved support vector machine (SVM) model, enhanced by the cuckoo algorithm (CA) (SVM-CA), for simulating monthly ET0.
• To compare the performance of the proposed SVM-CA model against established methods like genetic programming (GP), model tree (M5T), and adaptive neuro-fuzzy inference system (ANFIS).

Main Methods:

• Utilized meteorological data including maximum and minimum temperature, relative humidity, wind speed, and sunshine hours as inputs.
• Implemented the SVM-CA model for monthly ET0 simulation.
• Benchmarked SVM-CA against GP, M5T, and ANFIS using root mean square error (RMSE) and mean absolute error (MAE).

Main Results:

• The SVM-CA model demonstrated superior accuracy in simulating monthly ET0 compared to GP, M5T, and ANFIS.
• SVM-CA achieved reductions in RMSE of 5-15% (vs. GP), 12-21% (vs. M5T), and 7-15% (vs. ANFIS).
• SVM-CA showed reductions in MAE of 5-17% (vs. GP), 10-22% (vs. M5T), and 5-18% (vs. ANFIS).

Conclusions:

• The proposed SVM-CA model offers a highly accurate and reliable approach for simulating monthly ET0.
• SVM-CA presents a promising tool for enhancing agricultural water management strategies through precise ET0 prediction.