直接同化多种卫星辐射率资料对江淮暴雨预报影响研究

发布时间：2018-04-04 11:40

  本文选题：高光谱红外资料　切入点：卫星资料同化　出处：《南京信息工程大学》2016年硕士论文

【摘要】：近年来暴雨、冰雹等自然灾害频发,给人们生活带来不便,给社会经济带来了巨大的损失。高光谱红外传感器可提供高分辨率的大气垂直温度和湿度数据,引入该数据已证明可以很大程度的改善全球模式的精度。对于区域模式而言,同化高光谱红外资料仍然存在很多问题待解决。现阶段国内有很多将AIRS数据同化到区域模式中研究,而对于欧洲中心的高光谱红外IASI数据,国内研究较少,仅对海上台风有少数同化模拟研究。本文利用三维变分同化方法,结合中尺度WRF模式,探究IASI数据同化对区域模式暴雨模拟的改进效果。首先针对2014年6月一个月的预报结果,利用美国NMC方法统计研究区域的背景误差协方差。利用统计得到的背景误差协方差,针对2014年6月25～28的暴雨个例设计了两组试验。第一组试验为确定IASI资料具体的同化波段,对比分别同化常规观测数据、IASI温度探测波段、IASI湿度探测波段以及IASI所有波段,比较各组IASI试验同化结果。第二组试验根据第一组得到的结论同化IASI温度探测波段,并与同化AMSUA、MHS、 HIRS4、ATOVS试验进行对比,分析几种资料对江淮暴雨预报效果改进的影响。得到主要结论为：(1)针对模拟区域采用美国NMC方法统计背景误差协方差,可得到非平衡温度和假相对湿度是局地性很强的量,而流函数和非平衡速度势受边界层影响较大。(2)从第一部分试验可知在代价函数和梯度图上证明同化IASI温度探测波段试验更易达到收敛；对初始场改进更合理；降水落区和降水强度的模拟与实况降水最为接近,说明针对本次个例而言,同化温度探测波段比同化湿度探测波段更合理。因此针对本次个例主要选择同化IASI温度探测波段。(3)同化IASI温度探测波段试验与其他同化方案相比,进入同化系统内的数据点更多、对初始场改进效果明显、降水模拟和TS评分上结果也较好、误差分析中某些高度均方根误差小于其他方案。体现了IASI资料的应用价值。(4)对比同化AMSUA、MHS、ATOVS三个试验方案,发现在降水场模拟、误差分析上会出现同化AMSU A和MHS方案结果较好,而同化ATOVS (AMSU A+MHS+HIRS4)方案效果较差。原因很可能是：同化多个数据可能带来更大的观测误差,相互抵消正效应,反而导致结果变差。
[Abstract]:In recent years, heavy rain, hail and other natural disasters have brought inconvenience to people's life and great loss to social economy.Hyperspectral infrared sensor can provide high resolution atmospheric vertical temperature and humidity data, which has been proved to greatly improve the accuracy of the global model.For the regional model, there are still many problems to be solved in assimilation of hyperspectral infrared data.At present, there are many studies on assimilation of AIRS data into regional models in China, but there are few studies on hyperspectral infrared IASI data in the center of Europe, and only a few assimilation simulation studies on offshore typhoons.In this paper, using three-dimensional variational assimilation method and mesoscale WRF model, the improved effect of IASI data assimilation on regional model rainstorm simulation is explored.Based on the forecast results of June 2014, the background error covariance of the region is analyzed by using the NMC method in the United States.Based on the statistical background error covariance, two sets of experiments were designed for the rainstorm case on June 25, 2014.In order to determine the specific assimilation band of the IASI data, the first group of experiments compared the assimilation data of the conventional observation data, the temperature detection band and the humidity detection band and all the IASI bands, and compared the assimilation results of each group of IASI experiments.According to the conclusions of the first group, the second group of experiments assimilated the IASI temperature detection band, and compared with the assimilation IASI UAHS, HIRS4 / ATOVS test, and analyzed the influence of several kinds of data on the improvement of the forecast effect of Jianghuai rainstorm.The main conclusion is: 1) for the simulated region, the background error covariance is calculated by using the American NMC method, and the non-equilibrium temperature and the pseudo-relative humidity are found to be highly localized.However, the current function and the unbalanced velocity potential are greatly influenced by the boundary layer.) from the first part of the experiment, it is proved that the assimilation IASI temperature detection band experiment is more easy to converge, and the improvement of the initial field is more reasonable in the cost function and gradient diagram.The simulation of precipitation fall area and precipitation intensity is most close to the actual precipitation, which shows that the assimilation temperature detection band is more reasonable than the assimilation humidity detection band.Therefore, in this case, we mainly select assimilation IASI temperature detection band. 3) compared with other assimilation schemes, the assimilation IASI temperature detection band experiment has more data points entering the assimilation system, and the improvement effect on the initial field is obvious.The results of precipitation simulation and TS score are also good, and some RMS errors in error analysis are smaller than those in other schemes.The reason may be that assimilation of multiple data may result in greater observation errors and counteract positive effects, which may result in worse results.
【学位授予单位】：南京信息工程大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P457.6;P412.27
，

本文编号：1709807