Jalowska Anna M, Spero Tanya L
Office of Research and Development, U.S. Environmental Protection Agency, Durham, NC, USA.
J Geophys Res Atmos. 2019 Dec 27;124(24):13895-13913. doi: 10.1029/2019JD031584.
Extreme precipitation events influence watershed, agriculture, and urban management. The probability of extreme precipitation is estimated for storm water management using precipitation intensity-duration-frequency (PIDF) curves. This study explores developing PIDF curves from dynamically downscaled 36- and 12-km simulations using the Weather Research and Forecasting (WRF) model. Three modeled data sets are examined: 36-km WRF model forced with 2.5° (275-km) NCEP-DOE AMIP-II Reanalysis (R2); 36-km WRF model forced with 0.75° (80-km) ERA-Interim; and 12-km WRF model forced with ERA-Interim. The WRF outputs are verified against historical observations for three cities in the Eastern United States using a 23-year period (1988-2010). The 36-km WRF data set driven by R2 produced PIDF curves that were acceptable at the 12- to 24-hr durations, but those WRF data could not realistically simulate extremes represented by the high-intensity, short-duration precipitation events. Increasing the resolution of WRF's driving data from R2 to ERA-Interim did not improve WRF's representation of precipitation events. Using 12-km grid spacing enhances WRF's ability to reproduce PIDF curves developed from observations. Finer grid spacing dramatically improves the frequency and intensity of the 1- to 3-hr events and improves the 6- to 24-hr events. However, improvements with the 12-km WRF data did not apply equally to all study sites, suggesting further modifications to the WRF configuration and/or methodology are necessary. Although imperfect, the results here lend confidence to using modeled data to construct PIDF curves, which could be valuable for projecting changes to parameters used in urban and environmental planning.
极端降水事件会影响流域、农业和城市管理。利用降水强度-历时-频率（PIDF）曲线来估算用于雨水管理的极端降水概率。本研究探讨了使用天气研究和预报（WRF）模型从动态降尺度的36公里和12公里模拟数据中生成PIDF曲线。研究了三个模拟数据集：由2.5°（约275公里）的NCEP-DOE AMIP-II再分析（R2）驱动的36公里WRF模型；由0.75°（约80公里）的ERA-Interim驱动的36公里WRF模型；以及由ERA-Interim驱动的12公里WRF模型。利用1988年至2010年的23年时间，将WRF输出结果与美国东部三个城市的历史观测数据进行了验证。由R2驱动的36公里WRF数据集生成的PIDF曲线在12至24小时历时内是可以接受的，但这些WRF数据无法真实模拟高强度、短历时降水事件所代表的极端情况。将WRF驱动数据的分辨率从R2提高到ERA-Interim并没有改善WRF对降水事件的表现。使用12公里的网格间距增强了WRF再现从观测数据得出的PIDF曲线的能力。更精细的网格间距显著提高了1至3小时事件的频率和强度，并改善了6至24小时事件。然而，12公里WRF数据的改进并非对所有研究地点都同样适用，这表明有必要对WRF配置和/或方法进行进一步修改。尽管并不完美，但此处的结果为使用模拟数据构建PIDF曲线提供了信心，这对于预测城市和环境规划中使用的参数变化可能具有重要价值。
