Comparative prediction schemes using conventional and advanced statistical analysis to predict microbial water...

Area of Science:

• Environmental Science
• Agricultural Science
• Microbiology

Background:

• Accurate monitoring of indicator microorganisms is crucial for water quality management.
• Agricultural runoff, particularly from manure-treated land, poses a significant contamination risk.

Purpose of the Study:

• To compare Artificial Neural Networks (ANNs) and Multiple Regression Analysis (MRA) for predicting manure-borne microorganisms in agricultural runoff.
• To assess the impact of hydrological and environmental factors on microbial fate and transport.

Main Methods:

• Field rainfall simulation tests were conducted on agricultural plots treated with cattle or swine manure over a year.
• Microbial indicator concentrations were correlated with parameters like runoff volume, erosion, temperature, humidity, solar radiation, pH, EC, and turbidity.
• ANNs and MRA were employed to model and predict microbial concentrations.

Main Results:

• Artificial Neural Networks (ANNs) showed superior accuracy in predicting microbial concentrations compared to MRA.
• ANNs effectively modeled the nonlinear relationships between manure application and microbial transport.
• Key hydrological and environmental parameters significantly influenced microbial fate and transport.

Conclusions:

• ANNs offer a more robust approach for modeling microbial contamination in agricultural runoff.
• The findings support the use of ANNs for enhanced water quality monitoring and risk assessment in agricultural environments.
• Improved prediction models are essential for ensuring the safety of agricultural water resources.