Recovering the release history of a pollutant intrusion into a water supply system through a geostatistical approach

This work addresses the issue of recovering the release history of a pollutant injection in water distribution systems through the application of geostatistical tools. Geostatistical tools are here applied to the inverse problem of recovering the release history of a pollutant injected into a water‐distribution pipeline network. This application assesses the uncertainty in the concentration measurements and the release function by considering them as random variables characterized by their statistical moments. Performance of the geostatistical application to steady and unsteady flow conditions hinges on numerical modelling of the hydraulic behaviour of the water distribution system and evaluation of a transfer function that characterizes the probability of contaminant migration at control points in the network. A sensitivity analysis is carried out and it proves that the methodology is stable with respect to measurement errors and uncertainty in the network description (roughness coefficients and water demands). The analysis of error terms, the relative error of the released contaminant mass and the root mean square error of the recovered release function, demonstrates that the geostatistical methodology is able to characterize the release history of a contaminant in a water‐distribution pipeline system.