基于暴雨数值预报模式AREM,以2008年5月21日-7月30日为例,开展了以LAPS系统和GRAPES-3DVAR系统两种不同初值方案同化相同资料源(NCEP预报场、地面、探空资料)的AREM.LAPS和AREM.3DVAR试验,以探讨两种初值分析方案对降水模拟的影响。结果表明:(1)AREM-LAPS试验与AREM-3DVAR试验相比,各区域、各量级、各时效的降水预报在绝大多数情况下TS评分有较明显提高,特别是大雨、暴雨、大暴雨等强降水等级。(2)AREM-3DVAR试验可以大致模拟出与实况比较相近的平均降水量分布,但预报雨区范围偏小,强度偏弱;AREM.LAPS试验对此有较明显改善。(3)AREM—LAPS较AREM.3DVAR试验更好地模拟出了降水雨带的南北摆动及降水强度的变化。(4)AREM-LAPS试验较好地再现了我国西南地区东部、华南沿海、长江中下游.淮河流域及胶东半岛主要雨带区平均降水率逐日降水增强与减弱的过程,其强度也与实况大体相当;而AREM-3DVAR试验模拟的强度则明显偏弱,特别是我国西南地区东部。(5)对2008年夏季10余次典型降水过程的对比检验表明,AREM—LAPS对雨带范围、位置、强度的预报都好于AREM-3DVAR,特别是对于降水强度的改进尤为突出。
In this paper, based on heavy rain numerical forecast model AREM (Advanced Regional Eta Model), two different initialization schemes, LAPS and GRAPES-3DVAR, are used to conduct assimilation experiments, AREM-LAPS and AREM-3DVAR, with the same data (NCEP forecast field, surface data and radio-soundings) for the period from 21 May to 30 July 2008 for investigating the effect of the two initialization schemes on rainfall simulation. The result suggests as follows. (1) The forecast TS score by the AREM-LAPS is higher than that by the AREM-3DVAR for rainfall in all areas, at all valid time and with all intensity, especially for the heavy rain, rainstorms and extremely heavy rain. (2) The AREM-3DVAR can roughly simulate the average rainfall distribution, but the forecast area is smaller and rainfall intensity is weaker than the observation, while the AREM-LAPS significantly improves the forecast. (3) The AREM-LAPS gives a better forecast of a north-south wiggling of rainfall bands and rainfall intensity variation than the AREM-3DVAR. (4) The AREM-LAPS can give a better reproduction of the daily change in mean-rainfall rates of the main rain band, rainfall intensity changes in the eastern part of southwest China, the coastal area in South China, the middle- and lower-reach valleys of Yangtze River, the valleys of Huaihe river, and Shandong peninsula, with the rainfall intensity roughly comparable to the observation, while the rainfall intensity simulated by the AREM-3DVAR is clearly weaker than the observation, especially in the eastern part of southwest China. (5) The verification of comparisons between the AREM-LAPS and AREM-3DVAR for more than 10 typical rainfall processes in the 2008 summer indicates that the AREM-LAPS gives a much better forecast than the AREM-3DVAR in rain-band area, rainfall location and intensity, and in particular improves rainfall intensity forecast obviously.