Interaction Of Shallow Groundwater And Surface Ecological Environment In Tianjin City

There is a closed relationship between groundwater level in the shallow aquifer and surface ecological environment,so it is important for eco-environment construction to simulate and predict the groundwater level accurately.General speaking,the models for groundwater level predicting can be divided into deterministic models and stochastic models.The deterministic models are built based on some physical genesis,so they can predict long-series of groundwater level accurately;while the stochastic models are non-deterministic models without considering the interior structure of aquifers,and they determine the change rule of groundwater level based on the theory of probability and statistics,so they can predict short-series of groundwater level accurately.Firstly,the paper developed a refined groundwater flow model in the shallow aquifer with Groundwater Modeling System(GMS).In dealing with the recharge items,the quantity of irrigation return flow was calculated according to every actual irrigated areas;the rivers and the reservoirs were refined to secondary tributaries and small scale reservoirs,which made the recharge of the model more precise.In dealing with the discharge items,except the traditional discharge,the paper considered the interactions between the shallow aquifer and the deep.Then the model was calibrated in the four aspects: parameter identification,groundwater flow fields,groundwater head change over time and water balance.The result showed the precision of model was high,and it was accordant with the practical situation.Secondly,the paper built groundwater level prediction models by Multiple Linear Regression(MLR)and Quantile Regression(QR)method,and the models were applied to 5 sites of Tianjin city.Compared the predicted values in optimal quantile by QR model with the values by MLR model,the results showed that the related factors in 5 sites didn't follow the standard normal distribution and there were outliers in the precipitation,last month groundwater level factors,because of which the basic assumptions of MLR model couldn't be met,and caused errors,nevertheless the above condition had no effect on the QR model,it could describe the distribution of original data better,and had a higher precision in fitting extreme distribution data.Finally,the paper determined the vegetation distribution area and the suitable depth of groundwater level according to the distribution of vegetation in Tianjin city.The built deterministic model was applied to predict the water level dynamic variation in the vegetation distribution areas in 2020 and 2030 year.A plan was proposed to restore the vegetation of Tianjin in the paper,and the result showed the groundwater level in the vegetation distribution areas could be restored to the suitable values in the future.