Evaluation Of Parametric Inversion In Contaminant Reactive Transport Modeling

Groundwater pollution has been more critical as the petroleum and chemical industries developed rapidly. Kinds of organic components infiltrate into vadose zone and eventually reach the aquifer. The dissolved compounds are transported in groundwater with a larger scale than the NAPLs (non-aqueous phase liquids) in soils and aquifer, and also has a more direct influence on water quality. Therefore the study of contaminant transport and physical and chemical changes of dissolved components are of great importance on the contaminant remediation.It is worthy of considering the model reliability before the actual application of model, hence parameter estimation and model fit evaluation based on the reaction transport model are indispensable. In the inverse problem the random error was designed and added as the observation. Two weight schemes was taken to study the effect on estimation results, and it showed large discrepancy between two model scenes. The second scheme properly reflecting the observation error behaved better on the fit of simulation and observation. The conventional model fit graph showed evident defect whereas the residual analysis had more advantages to reveal the model deviation. The weighted residuals varied at random with simulated values and space independent variables. The statistics of error variance computed from objective function explained the consistence in model fit and the statistics used to calculate the weighting, and the model accuracy is reasonable.The parameter uncertainty is one reason that leads to model fit deviation with the observation, and carrying parameters sensitivity and correlation analysis is an effective way to understand the model internal system. The sensitivity of multi-species mole fraction are very small, while the microbial parameters are greatest and only the naphthalene’s rate transfer constant could match with it. This indicates that the transport and dispersion of tar oil plume was mainly controlled by the microbial degradation. Parameters correlation matrix shows the correlation of rate transfer constant and multi-species mole fraction. The eigenvector pertaining to the highest eigenvalue defines the direction of maximum insensitivity of the probability ellipsoid in adjustable parameter space. Three significant components defines the direction of the eigenvector which has higher eigenvalue, and it is an indication of three parameters correlation from rate transfer constant w1, w2 and multi-species mole fraction r2.Parameter uncertainty can be reduced by including more measurement data related with the offending parameters in the inversion process. The analysis emphasized the main role of microbial degradation in plume transportation and provided advices of choosing effective remediation technique for the decision maker.