With the advent of automatic sensors, detection, and data collection, for example, SCADA systems, it is now possible to collect a large number of time series of critical data. In urban water and sewer systems, monitoring stations can collect data on water quantity and quality, measuring, for example, dissolved oxygen, electric conductivity, pH, turbidity, among others. The motivation for such data collection usually is to analyze, if the systems are working in order. We present new analytical techniques in order to efficiently analyze such large quantities of data to answer questions of a forensic nature. Dozens of features are extracted from samples of time series data using linear predictive coding taps, Mel Frequency Cepstral coefficients, power spectral density, wavelet decomposition coefficients, and others. Classification of the set of time series is done using techniques that include both classical statistical techniques and techniques based on neural networks and fuzzy logic. We demonstrate these techniques with some real-world time series examples.