| In recent years,with the rapid population growth and the frequent social and economic activities,the consumption of groundwater resources has been accelerated.At the same time,various industrial and agricultural production and development have brought serious pollution to water bodies,and groundwater resources have been destroyed,resulting in hydrogeological disasters of groundwater levels decrease,land subsidence and seawater intrusion.It also leads to environmental problems of water quality deterioration and water sources pollution.Numerical models can effectively simulate and predict groundwater level and water quality.However,due to the strong uncertainty of the main hydrogeological parameters,i.e.,the hydraulic conductivity K and the longitudinal dispersilityα_L,model errors can not be ignored.Using data assimilation methods to assimilate observation data,hydrogeological parameters can be updated and the model errors can be reduced,which can provide references for the groundwater pollution prediction and groundwater monitoring well optimization in future.Ensemble Kalman Filtering(En KF)is an efficient data assimilation method,so it is widely applied in various study fields.In this study,source codes of MODFLOW and MT3DMS were used to establish the coupled model of MODFLOW-En KF and MODFLOW-MT3DMS-En KF.The visualization software of GMS was used to construct the groundwater flow numerical model and the solute transport numerical model in the Pinggu Plain district,obtaining the model input data for coupled model of MODFLOW-En KF and MODFLOW-MT3DMS-En KF.And then data assimilation systems of the groundwater flow model and solute transport model based on the En KF were developed in Pinggu distric,where hydrogeological parameters of K andα_L were inversed and updated and the model predictions accuracy were increased via using the observation data of hydraulic head and solute concentration.The influences of different factors on the En KF results were discussed.It was found that the larger ensemble size,the better the inversed results.And the ensemble size of 300 was selected in this study considering both the simulation accuracy and calculation cost.The K andα_L was inversed respectively via assimilating the groundwater solute concentration data,and the results were similar.This indicated that K andα_L were of the same importance and sensitivity to the groundwater solute transport model.Moreover,using concentration observation data to simultaneously invert the two parameters K andα_L shows a better data assimilation effect than inverting K orα_L separately.And the inversed results were better via assimilating both the groundwater hydraulic head and the solute concentration data than via assimilating the solute concentration data.Spatial distribution and observation frequency for the current monitoring wells in the Pinggu district were analyzed.The results showed that 58monitoring wells in Pinggu district were too many,resulting in data redundancy.When the number of the monitoring wells is reduced to 40,the data assimilation method is most efficient with least model error,and the data can be fully utilized,reducing the data redundancy and corresponding monitoring cost.In addition,it is recommended that the monitoring frequency should not be less than the current frequency of every quarter once. |
PDF Full Text Request