【作者】 王瑾；

【导师】 刘黎平；

【作者基本信息】 中国气象科学研究院 ， 大气物理与大气环境学， 2008， 博士

【摘要】 本论文使用贵州区域雷达网的多雷达拼图资料,以及504个防雹炮点的冰雹观测资料和其它资料(NCEP资料及DEM资料),通过分析研究贵州冰雹的雷达统计特征,对各种技术进行综合集成,建立有效的冰雹天气识别方法;并在此基础上,通过区域雷达资料同化技术的应用,建立基于强风暴数值预报的冰雹等强对流天气的临近预报方法,提供未来3小时内贵州区域冰雹的移动和发展的预报。论文的主要研究内容和研究成果如下:采用气候统计分析和基于GIS的数字地形分析、分区统计和图像分类方法,研究了贵州降雹的气候统计特征,并在此基础上研究冰雹分布与地形高程、坡向、坡度及地形切割深度的关系。研究表明,贵州降雹的空间分布与海拔高度、地形和下垫面性质等关系密切,具有明显的局地性特征。地形高程是影响贵州降雹分布的最主要地形影响因子。年平均降雹日数随地形高程增加呈增加趋势,在高程1000-1500m增加明显;微观地形因子如坡向和坡度对降雹日数的变异并没有显著性影响,但大范围的地势抬升及暖湿空气的迎风坡有利于降雹;地形切割深度并不是年平均降雹日数差异的显著性影响因子;纬度位置的不同,由于受暖湿空气影响程度不同及热力条件的差异,也是影响平均降雹日数差异的因子之一;根据3个影响因子建模获得的方程及贵州冰雹风险分区图,经统计检验和与历史乡镇降雹资料比较,具有较好的一致性。通过订立一定的规则,使用贵州504个防雹炮点的冰雹观测资料及贵州2005、2006年8次贵阳雷达站冰雹个例观测资料,使用“时间窗”方法间接地将风暴单体与降雹记录相联系,从而建立了冰雹算法校验数据库,然后通过对降雹校验数据库的统计分析,使用探测概率(POD)、虚警率(FAR)、临界成功指数(CSI)来检验冰雹探测算法。POSH的强冰雹探测算法的总体评估结果表明,30%的POSH强冰雹冰雹预警阈值在贵州地区获得最高的CSI评分,但这个阈值在每一次强冰雹的预警时也并未都获得最佳的结果。预警阈值选择模式(WTSM,即根据冻结层高度动态选择每一天的SHI的预警阈值)在不同地区气候状况下的差异,是导致缺省的POSH算法在贵州地区应用不佳的最主要的原因,这也说明对冰雹探测算法的局地性适用评估是非常必要的。通过对WTSM的调整,我们改进了原来的POSH算法,对它的重新评估结果表明,改进的POSH算法降低了识别的虚警率,对贵州地区的强冰雹识别效果还是相当好的。采用局部空间插值方法将雷达资料插值到规则的直角坐标网格点上,然后通过最大值法进行多雷达拼图;使用多雷达三维插值拼图产品,实现基于格点的垂直累计含水量、垂直累计含水量密度、强冰雹指数及强冰雹概率等强冰雹诊断因子,并加入冰雹地形影响因子,改进强冰雹概率算法。通过对一次发生在贵州西北部到中部一线的冰雹过程的成功诊断识别,说明了我们所开发的基于多雷达拼图的新的强冰雹诊断产品对识别贵州地区的强降雹是有比较好的效果的。以ARPS模式及其资料三维同化系统ARPS3DVAR和复杂云分析模块为研究平台,使用贵阳多普勒雷达的体扫观测资料,对一次强雷暴天气过程进行研究,分析了雷达反射率因子资料对云微物理量场的同化问题,并分析了雷达资料质量控制对同化的影响。在数值试验中使用了双层单向嵌套网格,在3km网格上设计了三个数值试验,分别使用不同的雷达反射率同化方案反演水凝物场。研究表明,数值试验中两个不同的云分析方案都能同化出合理的云水场和水凝物场,同时因为水凝物场的调整,位温场和垂直速度场也得到了很好的响应;与没有使用云分析方案的试验比较,使用了云分析方案同化雷达资料的试验,因为在初始时刻能够调整出一个合理的云微物理量场,抓住了对流风暴的主要特征,因此能够减少模式的热启动时间,更准确地模拟出初始时刻和短时的风暴的主要结构和演变特征,而没有使用云分析方案的试验在2到3小时后才调整出一个位置有较大偏差的水凝物场,因此,雷达反射率资料的同化在强对流风暴的模拟中起到了一个非常关键的作用。利用基于格点冰雹识别的研究成果,结合ARPS风暴数值模式的输出结果,提出基于风暴数值模式的冰雹临近预报方法,即用风暴数值预报的水物质场反演的反射率因子场作为冰雹的预报因子,并通过建立基于格点的强冰雹识别算法作为冰雹预报模型,从而对冰雹的落区及大小做出预报。与一般的冰雹预报模型相比,新的方法有以下特点:选取的冰雹预报因子物理意义更加明确,更加全面;建立的冰雹预报模型比较稳定;建立冰雹预报模型的过程相对简单。新的方法在一次强冰雹过程中得到了成功应用,在3h的临近预报中基本准确预报了强冰雹的落区位置。更多还原

【Abstract】 An effective hail detection algorithm suitable for Guizhou region has been established based on the analysis of the radar statistical characters of hail storms and the application of other comprehensive techniques by using 3D merged reflectivity of multiple radars over Guizhou ,hail events observation data of 504 hail prevention spots in Guizhou , and NCEP reanalysis data. A nowcasting technique for hail storm based on storm numerical model and radar data assimilation also has been established, which is used to forecast the moving and evolution of hail storm during 3 hours in Guizhou region. The main contents and conclusions are as follows:1. The climatic statistical characters of hail in Guizhou and the relationship between distribution of hail and some topographical factors, such as elevation, slope grade, slope aspect and terrain incision depth, has been studied by using climatic statistical analysis and some GIS techniques, such as digital terrain analysis, zonal statistics and image classification with historical hail records of 84 meteorological stations over 44 years in Guizhou province and the 1:1000000 resolution DEM data of china. It is shown, natural logarithm of mean annual hail days conforms to normality distribution .The elevation is the major topographical factor which influence the distribution of hail primarily, the annual mean hail days increase with the increase of elevation and it increase remarkably as the elevation increase to about 1000 -1500 meters. Micro topographical factors, such as slope grade and slope aspect are not remarkable factors to the difference of annual mean hail days, but topography rising over large area and windward slope of warm moist air is favorable to hail. Terrain incision depth is not remarkable factors to the difference of annual mean hail days also. Difference of latitude is also one of the factors which influence the difference of annual mean hail days. The model for annual mean hail days from the three remarkable factors and the map of hail hazard evaluation are credible via statistical test and comparison to historical hail reports over countryside spots.2. The evaluation databases for the WSR-88D hail detection algorithm have been built via the methodology defining specific conditions a storm cell must meet to be included in the evaluation by using hail observation data of 504 hail prevention spots in Guizhou and Doppler radar data of Guiyang during 8 of severe hail events from 2005 to 2006,after that, The algorithms were evaluated using the probability of detection (POD), false alarm ratio (FAR), and critical success index (CSI) statistics. It has shown that, (Probability of severe hail)POSH=30% got the highest CSI score in Guizhou region, but this threshold did not always get the best performance in these 8 of severe hail events. The difference of WTSM (Warning Threshold Selection Model) in different climatic region is the main reason why the default POSH algorithm got a bad performance. An improved WTSM would more accurately predict the optimum SHI threshold for each day. It would help ensure that the POSH threshold of 50% always corresponds to the largest possible CSI on each day. The re-evaluation of the improved POSH algorithm has shown that it has decreased the hail detection FAR, and get a higher performance of severe hail detection in Guizhou.3. Approach of Local space interpolation has been used to remap raw radar reflectivity fields onto a 3D Cartesian grid with high resolution, and maximum scheme has been used to combine multiple-radar reflectivity fields. Based on 3D merged reflectivity product, some severe hail diagnose factors, such as grid-based VIL, grid-based density of VIL, grid-based SHI (severe hail index) and grid-based POSH, have been calculated .The grid-based severe hail detection algorithm has been improved with the statistics of these severe hail diagnose factors and hail topographical factors. It is shown that the grid-based severe hail diagnose product would more accurately detect severe hail storm in a severe hail weather case which occurred in West-Northern and central Guizhou .4. For the purpose of researching the impact of assimilation of microphysical adjustments using reflectivity of Doppl er radar, ARPS model, ARPS3DVAR and a complex cloud analysis procedure are used to assimilate CINRAD/CD Doppler radar data of a severe storm weather case which occurred in Northern Guizhou province. The assimilation and predictions use a 3km grid nested inside a 15km one. There are three numerical experiments with different settings for retrieving cloud hydrometeors using reflectivity of Doppler radar in our studying. Results show that these two experiments with complex cloud analysis procedure can analyze a suitable distribution of cloud water and hydrometeors, and the distribution of potential temperature and vertical velocity response to the adjustment of hydrometeors perfectly. When compare to the EXPR_NOREF experiment without complex cloud analysis, because the CNTL-SMITH-REFONLY experiment which used the complex cloud analysis has analyzed a suitable field of cloud water and hydrometeors in model’s initial time, capture the major features of convective storm, so duration of spin up time has be decreased remarkably and the major structure and features of the storm in initial time and over 1h forecasting time can be simulated suitably, but the experiment without cloud analysis must cost 2 or 3 hours to analyze a field of hydrometeors which has error position. The assimilation of reflectivity of Doppler radar has acted as key role in the simulation of severe convective storm.5. Based on grid-based hail detection algorithm ,storm numerical model and radar data assimilation, A nowcasting technique for hail storm ,which using model radar reflectivity retrieving from hydrometeors as the hail forecasting factors and grid-based hail detection algorithm as the hail forecasting model, has been established, It can be used to forecasting the hail location and size. Comparing with other hail forecasting approaches, this new approach get more meaningful and more comprehensive forecasting factors, a more stable forecasting model and easy way to build the hail forecasting model. The new approach was successfully applied in a severe hail weather case, which forecasted accurately the position of severe hail storms during 3 hours from initial time.更多还原