Abstract: A squall line hit Beijing on Sept. 14, 2008 when thunder, light and strong winds was observed in the urban areas, however, the weather forecast for that day failed. The weather research and forecasting model (WRF) and the ARPS (advanced regional prediction system) data assimilation system (ADAS) are used to analyze the sensitivity of model horizontal resolution and initial condition contributed by different kinds of observations (FY2C satellite data, MICAPS data and GTS (global telecommunication system) data). Results show that there is no precipitation at all in Beijing in the simulation with a horizontal resolution of 15 km. While, for the simulation with a horizontal of 5 km, the precipitation can be simulated except that the area and intensity of the rainfall is different. There is no precipitation in the urban area in the control experiment without data assimilation. The simulated precipitation location is improved after using the ADAS data assimilation system. A MICAPS sounding in the Hetao area brings stronger northeast winds at lower level in the initial condition compared with control run in which the simulated squall line moves to the north of Beijing.

Key words: numerical model, data assimilation, numerical assimilation

摘要： 2008年9月14日一个飑线袭击北京, 城区发生了短时雷雨和大风天气, 而当时的天气预报失败。利用WRF (weather research and forecasting)模式和ARPS (advanced regional prediction system)模式的资料分析系统ADAS (ARPS data analysis system), 研究了模式水平分辨率, 不同观测资料(风云2C卫星资料、MICAPS数据、GTS (global telecommunication system)数据)同化所产生的初始场对这次飑线过程模拟的敏感性。结果表明, 15 km模式分辨率不能模拟出北京的降雨过程, 而5 km水平分辨率模式均能模拟出北京的降雨过程, 只是降雨落区和强度有所不同。不用资料同化的对照试验不能模拟出城区主要降水区。使用资料同化系统ADAS后, 降水落区均有改善。MICAPS资料中河套地区一个探空站的加入, 使得初值中河套地区的东北风比控制试验强, 模拟的飑线比控制试验偏南, 和实况更加接近。

关键词: 数值模式, 资料同化, 数值模拟