Development of artificial intelligence models for well groundwater quality simulation: Different modeling scenarios

Area of Science:

• Earth Science
• Environmental Science
• Artificial Intelligence

Background:

• Groundwater is a vital freshwater source, particularly in arid regions facing water scarcity.
• Assessing groundwater quality is critical for agricultural, drinking, and industrial uses.
• Traditional geostatistical methods offer insights but advanced AI can improve spatial quality predictions.

Purpose of the Study:

• To evaluate the effectiveness of various Artificial Intelligence (AI) models in simulating key groundwater quality parameters.
• To compare the predictive performance of different AI models for electrical conductivity (EC), sodium adsorption ratio (SAR), total dissolved solids (TDS), and sulfate (SO4).
• To identify the most accurate AI model for groundwater quality assessment in the Tabriz Plain, Iran.

Main Methods:

• Utilized a dataset from 90 wells in Tabriz Plain, Iran.
• Applied and assessed multiple AI models, including a hybrid Support Vector Machine-Firefly Algorithm (SVM-FFA).
• Employed k-fold cross-validation and two simulation scenarios incorporating geographical information and other quality parameters.

Main Results:

• AI models demonstrated strong capabilities in modeling groundwater quality variables.
• The hybrid SVM-FFA model exhibited superior predictability across both investigated scenarios.
• The study confirmed the potential of AI for accurate groundwater quality assessment.

Conclusions:

• AI models offer a robust and accurate alternative to traditional methods for groundwater quality modeling.
• The developed SVM-FFA model provides a reliable computational tool for groundwater monitoring and management.
• This research highlights the value of AI in addressing complex geo-science challenges related to water resources.