Predicting monthly evaporation from dam reservoirs using LS-SVR and ANFIS optimized by Harris hawks optimization...

Area of Science:

• Hydrology
• Environmental Science
• Data Science

Background:

• Evaporation measurement is vital for hydrological studies but costly.
• Machine learning offers a cost-effective solution for evaporation prediction using limited data.

Purpose of the Study:

• To evaluate and compare machine learning models for predicting monthly evaporation from dam reservoirs.
• To investigate the performance of adaptive neuro-fuzzy inference system (ANFIS) and least-squares support vector regression (LS-SVR).
• To enhance prediction accuracy using the Harris Hawks Optimization (HHO) algorithm to optimize ANFIS parameters.

Main Methods:

• Utilized monthly hydrological and meteorological data from Doroudzan dam (October 1999 - September 2020).
• Compared ANFIS and LS-SVR models with various input variable combinations.
• Optimized ANFIS using the HHO algorithm to create a hybrid ANFIS-HHO model.

Main Results:

• LS-SVR (RMSE=2.77, MAPE=2.48, NSE=0.93) outperformed the basic ANFIS model.
• The hybrid ANFIS-HHO model achieved superior evaporation prediction (RMSE=2.35, MAPE=1.55, NSE=0.95).
• Taylor's diagram confirmed the enhanced performance of the ANFIS-HHO model over LS-SVR and ANFIS.

Conclusions:

• The hybrid ANFIS-HHO model provides a highly accurate and reliable method for predicting dam reservoir evaporation.
• The proposed methodology is effective for evaporation prediction influenced by various dam-related variables.
• Excluding lake area from input variables yielded the best prediction results across models.