Abstract

Under changing environments, conventional methods based on stationarity assumption can lead to unreliable estimates of design rainfall quantiles. In this study, the trend of six extreme precipitation indices in Beijing-Tianjin-Hebei region was analysed using different methods. Nonstationary models were established for annual maximum one-day precipitation (RX1d) and annual total precipitation when daily precipitation greater than 95% of daily precipitation in a year (R95p) exhibiting significant trend, while stationary models were established for the two indices showing no significant trend. Finally, the nonstationary and stationary design precipitation were estimated using the average design life level method and the stationary method, respectively. The results indicated that: for some stations in the east and south, a significant decreasing trend was observed for RX1d and R95p. This resulted in the overestimation of their design precipitation using the stationary methods and altering the spatial distribution of precipitation extremes for future period near these stations.

摘要

在不断变化的环境下，基于平稳性假设的常规方法可能导致设计降雨分位数的估计不可靠。本研究采用不同方法分析了京津冀地区6个极端降水指数的变化趋势。年最大单日降水量（RX1d）和日降水量大于全年日降水量95%的年总降水量（R95p）呈现显着趋势建立非平稳模型，而对两个指标均不显着建立平稳模型趋势。最后，分别使用平均设计寿命水平法和平稳法估算了非平稳和平稳的设计降水量。结果表明：对于东部和南部的一些站点，RX1d 和 R95p 呈显着下降趋势。这导致使用固定方法高估了他们的设计降水，并改变了这些站附近未来时期降水极值的空间分布。