Prediction of water quality in Jordanian dams using data mining algorithms

Area of Science:

• Environmental Science
• Water Resource Management
• Data Mining Applications

Background:

• Effective water management is crucial in arid regions.
• Assessing and predicting water quality is vital for sustainable resource use.
• Jordanian dams face challenges in maintaining water quality.

Purpose of the Study:

• To assess and predict water quality in three Jordanian dams.
• To apply advanced data mining techniques for water quality index (WQI) prediction.
• To understand seasonal and annual water quality variations.

Main Methods:

• Collected physical, chemical, and biological water quality parameters over four years.
• Calculated the Weighted Arithmetic Water Quality Index (WA-WQI).
• Employed data mining algorithms: generalized linear models, decision trees, random forests, gradient-boosted trees, and Support Vector Machine (SVM).

Main Results:

• Significant water quality fluctuations were observed, influenced by pH, conductivity, nutrients, and microbial contamination.
• Support Vector Machine (SVM) demonstrated superior performance in predicting WQI compared to other models.
• The study identified key drivers affecting water quality in the studied dams.

Conclusions:

• Continuous monitoring and predictive modeling are essential for effective water resource management.
• SVM is a highly effective tool for water quality prediction in arid environments.
• This research provides a benchmark and predictive framework for water management in Jordan and similar arid regions.