GROUNDWATER MODELING - A CASE STUDY TINKER AIR FORCE BASE, OKLAHOMA

摘要

As a part of site remediation, groundwater models are frequently used as a basis for whether remediation is necessary and, if so, to obtain preliminary remediation goals. Actual field data are used as model input, and the model estimates future contaminant plume migration. Often, plume migration thirty years into the future is modeled. Typically, a large numerical model such as MODFLOW is used to define the underlying hydrogeologic conditions. In order to generate an optimal model, the correct type and an adequate amount of input data are required. In the majority of cases, these data are not available. A typical test for the correctness of the model is model calibration. By manipulating input parameters, the modeler attempts to make the model output fit the existing field data. If the fit is within certain statistical limits, the model is said to 'calibrate' which is intended to indicate that the model output is correct. The caveat to an elaborate deterministic model is that calibration can be achieved, and is often necessitated, by inaccurate and unrealistic input data or output data. That is, these models are often so complex that errors in one area of model input can be offset by errors in other variables so that calibration can be achieved with a wholly inappropriate model configuration. Ideally, verification with an independent data set should follow calibration. Too often however, these large, complex numerical models require so much input information that an additional verification step can not feasibly be attempted. At Tinker AFB, a complex MODFLOW model was generated to determine hydraulic heads and associated contaminant velocities for the northwest comer of the base. The model output was then used as input to a fate and transport model in order to simulate 30-year plume migration. The combined model results showed that for some contaminants, the plumes would be migrating well off base in 30 years, which could initiate a costly interim corrective action. Investigation of model input parameters indicated that unrealistically large hydraulic conductivities of 244 ft/day were needed in order to achieve model calibration. Fundamental statistical, hydraulic and geostatistical analyses later supported the hypothesis that the input parameters were fiawed and that the calibrated model produced excessive contaminant transport velocities. All too often, modeling results are submitted to the regulators without knowledge of many of the model parameters. Because these model results will likely form the basis for corrective actions at a site, incorrect model results can initiate unnecessary corrective actions. It is important to take time to investigate the model parameters as well as the type of model that is being used to see if they seem realistic given the known hydrogeology of the site.