Simulating reference crop evapotranspiration with different climate data inputs using Gaussian exponential model.

Area of Science:

• Agricultural Science
• Hydrology
• Machine Learning

Background:

• Accurate reference crop evapotranspiration (ET₀) data is crucial for efficient agricultural water management.
• Limited climatic data poses a challenge for precise ET₀ estimation.

Purpose of the Study:

• To develop and validate a novel Gaussian exponential model (GEM) for predicting ET₀ using restricted climatic variables.
• To compare the performance of GEM against M5 model tree (M5T), extreme learning machine (ELM), and boosted trees (BT) models.

Main Methods:

• Utilized daily meteorological data from four stations in Northeast China (1997-2016).
• Developed and validated GEM, M5T, ELM, and BT models using various input combinations.
• Evaluated model performance using metrics like RMSE, RRMSE, E<0xE2><0x82><0x99><0xE2><0x82><0x9B>, R², MAE, and GPI.

Main Results:

• Models incorporating solar radiation and relative humidity showed significantly higher accuracy.
• GEM demonstrated superior accuracy compared to M5T, ELM, and BT across different scenarios.
• GEM achieved higher accuracy under local scenarios versus regional scenarios, indicated by improved error metrics.
• GEM with solar radiation input achieved the highest accuracy (GPI = 1.876).

Conclusions:

• The Gaussian exponential model (GEM) is a highly accurate and consistent tool for simulating reference crop evapotranspiration (ET₀).
• GEM, particularly with solar radiation data, is recommended for ET₀ estimation in Northeast China.
• The developed model offers a valuable approach for agricultural water management under data constraints.