Abstract

The characteristics of urban pluvial flooding are altering all over the world due to environmental change. In this paper, generalized additive models for location, scale and shape (GAMLSS) was employed to analyze nonstationary frequency of extreme precipitation at subdaily scale. Nonstationary precipitation return periods were estimated using the expected waiting time (EWT) interpretation. A model coupling SWMM and MIKE21 was established, and calibrated and verified by three historical urban floods. Then, it was utilized to simulate urban floods under stationary and nonstationary rainfall conditions with different return periods. The simulated results illustrated that rainfall depth under nonstationarity was greater than that of stationarity when return period was less than 10 years, but the results reversed when the return period was over 20 years. The main variation of rainfall depth occurred within 6 h. The deviation of the maximum water depth was less than 10% for five return levels, and the difference in the longest inundation lengths was 1.2 h for 50-year return period under two assumptions. It may indicate that slight differences of urban flooding were detected between stationary and nonstationary conditions in the study region, which suggested a further study about urban flood under nonstationarity.

中文翻译：

---

基于SWMM和MIKE21耦合的降雨非平稳性对城市洪水的影响

摘要

由于环境的变化，世界各地的城市洪水泛滥的特征正在改变。本文采用位置，尺度和形状的广义加性模型（GAMLSS）来分析日尺度下极端降水的非平稳频率。使用预期的等待时间（EWT）解释来估算非平稳降水的返回期。建立了将SWMMH和MIKE21耦合的模型，并通过三个历史城市洪水对其进行了校准和验证。然后，它被用来模拟具有不同返回期的固定和非固定降雨条件下的城市洪水。模拟结果表明，非平稳期的降雨深度大于回归期小于10年的平稳期；但是当返回期限超过20年时，结果却相反。降雨深度的主要变化发生在6小时之内。在两个假设下，五个返回水位的最大水深偏差小于10％，并且对于50年返回期的最长淹没长度差异为1.2 h。这可能表明在研究区域内，在固定和非固定条件下检测到的城市洪水存在细微差异，这表明需要对非固定条件下的城市洪水进行进一步的研究。