Analysis of multispectral imagery and modeling contaminant transport

A significant concern in the monitoring of hazardous waste is the potential for contaminants to migrate into locations where their presence poses greater environmental risks. The transport modeling performed in this study demonstrates the joint use of remotely sensed multispectral imagery and mathematical modeling to assess the surface migration of contaminants. KINEROS, an event-driven model of surface runoff and sediment transport, was used to assess uranium transport for various rain events. The model inputs include parameters related to the size and slope of watershed components, vegetation, and soil conditions. One distinct set of model inputs was derived from remotely sensed imagery data and another from site-specific knowledge. To derive the parameters of the KINEROS model from remotely sensed data, classification analysis was performed on IKONOS four-band multispectral imagery of the watershed. A system known as GENIE, developed by Los Alamos National Laboratory, employs genetic algorithms to evolve classifiers based on small, user-selected training samples. The classification analysis derived by employing GENIE provided insight into the correct KINEROS parameters for various sub-elements of the watershed. The model results offer valuable information about portions of the watershed that contributed the most to contaminant transport. These methods are applicable to numerous sites where possible transport of waste materials poses an environmental risk. Because the approach rests on the analysis of remote sensing data, the techniques can be used to monitor inaccessible waste sites, as well as reduce the amount of data that would need to be collected for model calibration.