中文摘要：

分析咸潮上溯的规律以及对河口含氯度的模拟预测对河口地区供水安全具有重要的意义。以珠江三角洲磨刀门河口为例,上游流量和下游潮位作为边界条件,建立基于BP神经网络的含氯度变化模拟模型,通过对模型参数进行率定与验证,最终得出模拟结果与实测含氯度数据较为一致,河口含氯度与上游径流量和下游潮位之间的映射关系良好。基于此,研究上游径流量变化对含氯度的影响及不同来水频率和潮位设计方案下的含氯度变化。结果表明：当上游水电站采用降序泄水时可有效地抑制河口咸潮上溯;同一设计潮位方案下,特枯来水年份（97%来水频率）平岗泵站日平均含氯度较正常来水年份（90%来水频率）平均增幅达到3.3倍。

英文摘要：

Salt water intrusion of estuary not only affects agricultural development, but also has a serious impact on terrestrial freshwater re- sources. Thus, it is important to analyze the characteristics of salt water intrusion and simulate the estuarine chlorinity for water supply safety in estuary areas. Take the Modaomen estuary for instance, runoff and tide level are considered as boundary conditions, and a simulated model about the variation of chlorinity is then established based on BP neural networks. Through the calibration and validation of the modeFs param- eters, we can know that the simulated result is consistent with the measured value. It shows that the mapping relation between estuarine chlo- rinity, runoff, and the tide level is good. Based on the model, we simulate the change of chlorinity responding to the different design schemes of runoff and the tide level. The result shows that the descending outflow of the reservoirs can better inhibit the invasion of chlorinity in estuar- ies. And under the same designed tide level scheme, daily average chlorinity of Pinggang Station in the drought years is 3.3 times larger than that of the normal years.