Long-term water demand forecasting using artificial intelligence models in the Tuojiang River basin, China

Area of Science:

• Environmental Science
• Hydrology
• Artificial Intelligence

Background:

• Accurate water demand forecasting is essential for strategic planning and sustainable use of finite water resources.
• Effective water resource management underpins regional socio-economic development.

Purpose of the Study:

• To compare the applicability of various artificial intelligence (AI) models for long-term water demand forecasting across different water use sectors.
• To identify the most effective AI model for predicting future water demand in the Tuojiang River basin.

Main Methods:

• Five AI models were evaluated: Genetic Algorithm optimized Back Propagation Neural Network (GA-BP), Extreme Learning Machine (ELM), Gaussian Process Regression (GPR), Support Vector Regression (SVR), and Random Forest (RF).
• Models predicted water demand for agricultural, industrial, domestic, and ecological sectors using data from 2005-2020.
• Performance was assessed using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE).

Main Results:

• Gaussian Process Regression (GPR) showed superior performance for agricultural (R2=0.9811), domestic (R2=0.9338), and ecological (R2=0.9142) water demand.
• Genetic Algorithm optimized Back Propagation Neural Network (GA-BP) performed best for industrial water demand (R2=0.8580).

Conclusions:

• The study identified optimal AI models for long-term water demand forecasting in different sectors.
• These validated models offer a valuable tool for enhancing sustainable water resource management and planning.