Predicting bacterial transport through saturated porous media using an automated machine learning model

Escherichia coli, as an indicator of fecal contamination, can move from manure-amended soil to groundwater under rainfall or irrigation events. Predicting its vertical transport in the subsurface is essential for the development of engineering solutions to reduce the risk of microbiological contamination. In this study, we collected 302 datasets from 39 published papers addressing E. coli transport through saturated porous media and trained an automated machine learning model (H2O AutoML) to predict bacterial transport. Bacterial concentration, porous medium type, particle size, ionic strength, pore water velocity, and column length were used as input variables while the first-order attachment coefficient and spatial removal rate were set as target variables. The six input variables have low correlations with the target variables, namely, they cannot predict target variables independently. However, with the automated machine learning model, input variables can effectively predict the target variables. Among 20 candidate models, Gradient Boosting Machine showed the best performance. Among the six input variables, pore water velocity, ionic strength, particle size, and column length were more important than bacterial concentration and porous medium type. This method of using historical literature data to train automated machine learning models provides a new avenue for predicting the transport of other contaminants in the environment.