Sequential characterization of contaminant plumes\ud using designed monitoring network

In many practical field problems, it may not be possible to identify the actual characteristics (location, magnitude and duration of contamination) of the groundwater contaminant sources. Most of the time, very sparse information regarding spatiotemporal contaminant concentration is available initially, which is inadequate for reliable identification and simulation of the contaminant plume. Simulation of the contaminant plume is necessary to predict the future distribution of the\ud contaminant in the groundwater aquifer. It is also essential for developing an efficient contamination monitoring strategy. To address this practical problem of data inadequacy, an interactive methodology is proposed, incorporating sequential design of optimal monitoring\ud networks. These sequentially developed and implemented monitoring networks provide feedback information on measured concentrations. These measurement information help in progressively improving the prediction of the contaminant plume, starting with very sparse initial information about the contaminant sources and spatial distribution of concentration.\ud \ud A Simulated Annealing (SA) based methodology is developed which utilizes feedback information to sequentially characterize the contaminant plume when adequate initial information regarding the contaminant sources is not available, and the contaminant sources are unknown. It is demonstrated that such an approach is potentially applicable for reliable prediction of contaminant plume\ud movement in groundwater aquifers. The optimal monitoring network design is based on the objective function of minimizing the contaminant mass estimation error. It is shown that the proposed methodology of using feedback information on actual field measurement values from\ud sequentially designed monitoring networks is capable of predicting the contamination plume movement in groundwater aquifers, even when very little initial measurement data are available. This proposed methodology will be very useful for accurate prediction of the contaminant movement in an initially unmonitored aquifer