中文摘要：

针对水质时间序列一般具有结构复杂性和演变过程非平稳性的特征,提出将马尔可夫状态切换-自回归模型（Markov Switching-Auto Regressive Model）用于水质时序的随机模拟和水环境系统的风险分析,并探讨了运用MS-AR模型进行水质时序随机模拟与分析的关键步骤。最后运用该模型和蒙特卡罗（Monte Carlo）方法,对天津果河某断面的总磷浓度时序（1999—2007年）进行了随机模拟和污染风险评价。结果表明,MS-AR模型有效识别出了该水质时序在演变过程中的不同结构模式,模拟出的总磷浓度序列能从总体上反映实测总磷时序的演变特征和动态变化趋势,克服了利用单纯统计理论进行水环境风险分析无外延性的不足。

英文摘要：

Focusing on the non-stationary and structural complexity of water quality time series, the Markov Switching-Auto Regressive Model （MS-AR） was introduced for water quality time series random simulation, and then the main procedures of environmental risk assessment for water system by using MS-AR model was discussed. Finally, by means of this model and Monte Carlo method, the total phosphorus （TP） concentration time series in a river section in Guohe River, Tianjin, was randomly simulated, and the risk of TP exceeding national standard limits was assessment. The result shows that the MS-AR model can identify two evolution patterns of the series, the simulated series reflect the evolution features and the changing trend of the actual time series, which overcame the disadvantage of extensionality and prediction capability of using statistical method only to the environmental risk analysis of water system.