Evaluation of groundwater withdrawals in Tongzhou District, Beijing, China

The accurate estimation of groundwater withdrawals influences the evaluation and management of groundwater resources worldwide and the amount of groundwater withdrawals has an important impact on the ecological environment. However, many uncertainties associated with measuring groundwater withdrawals, such as the water meter method and quota method, have resulted in considerable errors in the estimations. In this study, two methods—the water balance method and back-propagation artificial neural network (BP-ANN) method—were both employed to evaluate the accuracy of groundwater withdrawals from official data. A case study was selected in the Tongzhou District of Beijing, the sub-center of Beijing, China. Based on the data of the groundwater level, precipitation, and official groundwater withdrawals from 2000 to 2014, the accuracies of the official data on groundwater withdrawals were then evaluated using the two aforementioned methods. The results revealed that the groundwater withdrawals estimations from the official data were more inaccurate than those from the water balance method, except in 2004. The water balance method is only used at a yearly time scale, whereas the BP-ANN method can be applied at both yearly and monthly time scales. Reliability analysis of the estimations revealed that the proposed two methods can effectively estimate groundwater withdrawals and can serve as an effective tool for groundwater resources management.