Comparative prediction schemes using conventional and advanced statistical analysis to predict microbial water quality in runoff from manured fields

Abstract

Accurate estimations of indicator microorganisms’ concentrations are necessary to properly monitor water quality and manage contamination from agricultural land runoffs. In this study, Artificial Neural Networks (ANNs) and Multiple Regression Analysis (MRA) statistical methods were compared for accuracy in the prediction of manure-borne microorganisms’ concentrations in runoffs from agricultural plots (0.75 m × 2 m) treated with cattle or swine manure. Field rainfall simulation tests were initiated on days 4, 32, 62, 123, and 354 between June 2002 and May 2003. Each rainfall event produced 35 mm rainfall for 30 min at the intensity of 70 mm hr⁻¹ at 24-intervals. Concentrations of microbial indicators were correlated with hydrological and environmental water quality parameters including water runoff, erosion, air temperature, relative humidity, solar radiation, pH, electric conductivity (EC) and turbidity to determine their impacts on microbial fate and transport. ANNs demonstrated a better ability to model the nonlinearity of land application of manure to ensure the safety of agricultural water environments.

Keywords:

• Artificial Neural Networks (ANNs)
• Multiple Regression Analysis (MRA)
• microbial fate and transport
• manure treatment

Acknowledgment

Many thanks are due to USDA-ARS scientists, Dr. John Gilley and Dr. Bahman Eghball (deceased), who provided valuable data to significantly contribute to this study. Mention of trade names or commercial products in this article is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U. S. Department of Agriculture.