Self-Organizing Linear Output (SOLO) Approach for Managing Total Coliform Indicator Bacteria on California Beaches

In this study, one of the artificial neural networks, Self-Organizing Linear Output (SOLO), was used to predict levels of indicator bacteria at Newport Bay in Newport, Beach, California, USA. The approach over-estimated several observations which showed extraordinary low concentrations compared to others. Average of observations was 6351 CFU/100mL and that of error value for model validations was only 176 CFU/100mL, about 3% of observation average without few points which showed extraordinary low concentrations. The results of this study showed that the approach was very effective for predicting concentrations of indicator bacteria. The approach was carried out for monthly average prediction because of limited dataset. The study could be extended for finer time scale prediction, such as weekly or daily prediction, when more measurements are available.