The feasibility of monitoring open-channel water systems as an early warning of the accidental or intentional release of biological agents was investigated. Critical steps in this study included (i) evaluation of the quantity of pathogens that would be released into the sewer system, (ii) how these organisms would be distributed in an open-channel system (accounting for dilution and dispersion), and (iii) how well they could be predicted at downstream locations. We developed and examined prediction models using computational tools such as CFD (Computational Fluid Dynamics) and ANNs (Artificial Neural Networks) for water collection systems though analyses of the collected data. The models were designed (i) to forecast microbial dispersion patterns in each system, (ii) to estimate dispersion time, and (iii) to recommend detection methods, sampling frequencies, and sampling locations. Based on a series of field experiments, those computational models which proved effective were designed to provide us with an impetus to establish an optimization technique for real-world situations. Field experiments and numerical simulation data were essential to evaluate the validity of the developed model. The use of ANNs for spatial and temporal identification of biological agents was conducted based on the particular characteristics resulting from pH, turbidity, and conductivity data corresponding to E. coli concentration over time. Overall, the simulation results for the two specific purposes of using ANNs, parameter estimation and feature classification, were highly satisfied (R2 = 0.77-0.96). It was concluded that ANNs could effectively be used for multiple tasks, such as prediction of the dispersion patterns of E. coli using its surrogates. In addition, various characteristics of the time-series concentration of E. coli, flow rate, inlet position, distance from an outlet, etc., were well considered in order to classify the release location and concentration.