基于雷达组网拼图和XGBoost的雷达定量降水估测

针对传统方法采用天气雷达进行强降水的定量估测存在较大偏差问题,论文以1 h累计雨量为估测对象,基于雷达组网拼图资料,采用XGBoost(eXtreme Gradient Boosting)算法,建立新的雷达估测降水模型。该模型设计以前1 h的雷达组合反射率因子作为输入,进一步采用若干个剔除异常样本的策略有效清除建模样本中的部分噪声,更好地构建了雷达组合反射率与估测对象之间的非线性映射关系。在32万个独立检验样本的估测结果中,其均方根误差(RMSE)为6.04 mm、平均绝对误差(MAE)为3.50 mm、预报偏差(BIAS)为1.05;与目前业务系统上使用的ZR(300,1.4)关系方法相比,前者的RMSE和MAE分别下降了20.6%和10.3%,而BIAS指标则显示后者对降水量级的估测明显低估。进一步对小时雨强大于10 mm样本的统计结果表明,新方案的RMSE、MAE以及TS评分均大幅优于ZR(300,1.4)关系方法,可进行实际业务应用。     

天气雷达联防报自动编制和降水回波图打印系统

详细介绍了系统的功能和技术问题，系统是对天气雷达数字化终端的再开发，它由天气雷达联防报自动编制和降水回波图打印两部分组成。系统用８０８８宏汇编语言编程。使用这个系统显著提高了天气雷达业务工作的自动化程度和精确度。     

用C波段双线偏振雷达研究冰雹云

本文计算了不同相态、不同形状的降水粒子对5.6cm,10.7cm雷达波的散射特性,并利用降水粒子散射特性及云雨雷达回波的Z_(DR)特征并结合地面降雹情况,分析并解释了1990年8月9日和8月30日甘肃省平凉两次雹暴过程的雷达RHI资料.我们推测:Z_(DR)值为负值的区域相应于降雹区,ZDR负值与大扁冰雹和小锥状冰雹关系密切;而雨区Z_(DR)具有较大正值,雨区大于5dB的Z_(DR)由直径D>0.5cm的雨滴所致.利用冰雹区和雨区Z_(DR)信息的不同可能识别降水粒子的相态.C波段双线偏振雷达和10cm波长雷达相比有独到之处:雨区Z_(DR)值较大,而冰雹区Z_(DR)值又较小,这更有利于研究云雨相态和空间结构.可以预期双线偏振雷达对防雹工作及云物理研究工作是有很大帮助的.     

用小波变换提取雷达回波图像中的辐合特征线

采用小波变换的方法对滨州雷达资料的反射率图像进行处理,得到特征线增强后的图像,将其输入自动预报系统实现特征识别,可为雷暴预报提供初步资料。     

CFL 03型风廓线雷达自动标校方法研究

风廓线雷达已经广泛应用于国民经济的许多领域,现在对于其应用已不仅仅局限于风场资料,对回波强度、信噪比、速度谱宽等信息的使用也越来越多。因此,对其开展标校工作就显得非常重要。以CFL-03型风廓线雷达为例,介绍了一种风廓线雷达的自动标校技术,从技术方案设计到信号流程、系统硬件组成,以及对其实现方法进行了研究,并用机外仪器对设备进行了测试,将测试结果和自动标校结果进行了对比,对比结果证明该自动标校系统的精度达到了设计目的,能够满足雷达业务运行要求。     

CINRAD/SA雷达日常维护及故障诊断方法

CINRAD雷达系统，具有完善的自动定标系统和可靠的故障自诊断系统，利用其“故障报警信息”和“RDA性能参数”，能够完成该雷达系统的日常维护任务；参照“RDA适配数据”，再配合简单的测量工具，可以实现故障的隔离和定位。     

Analysis of the Initial Stage Intracloud Lightning Using the Pulse Location Technique Based on the Fast Electric Field Change

Academy of Sciences,Beijing 100029 2 Lanzhou University,Lanzhou 730000 3 Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences,Lanzhou 730000 4 Nanjing University of Information Science＆Technology,Nanjing 210044
Using both fast and slow electric field change sensors and field mill,multi-station observations on lightning flashes over China inland plateau areas were conducted during the summer of 2004.All of the stations were synchronized by GPS with a time-resolution of±50 ns.Using the different time of arrival（DTOA）and based on the fast electric field change sensor,a lightning radiation location technique was developed.Radiation pulses in the initial stages of five intracloud（IC）lightning discharges which occurred on 20 August were analyzed.The results indicate that the technique developed could effectively locate the lightning radiation sources.Furthermore,the lightning discharges were compared with the Doppler radar data.The results show that the radiation sources were well associated with the storm development.When the storm was at the mature stage with an echo top of about 10.9 km,the radiation sources ranged from 5.2 to 8.3 km above mean sea level;when the storm gradually became weaker,with echo top of about 7.9 km,the radiation sources ranged from 3.0 to 5.9 km.In particular,one of the IC lightning discharges ranged from 3.0 to 4.9 km during the dissipation stage of the storm.The results also indicate that the radiation sources were closely associated with the high reflectivity region（25–50 dBZ）of the storm,which,to some extent,demonstrates the reliability of the location results,thereby showing that multi-station observations of fast electric field change sensors could be a useful tool for monitoring the storm development.Location errors from radiation sources were also compared with the radar data and the results of a simple simulation.It was found that the errors were getting smaller when the radiation sources approached the center of the detection network, and vice versa.Compared with the limited experimental observations,the simulation results were found capable of effectively reflecting the location errors.     

在雷达—雨量计联合探测系统中距离加权和时间外推校准技术的研究

本文用距离加权方法校准天气雷达估算降水强度和区域降水量。     

Analysis of mesoscale convective systems in Catalonia using meteorological radar for the period 1996–2000

The aim of this paper is to show a climatology of Mesoscale Convective Systems (MCS) in the NE of the Iberian Peninsula, on the basis of meteorological radar observations. Special attention was paid to those cases that have produced heavy rainfalls during the period 1996–2000. Identification of the MCS was undertaken using two procedures. Firstly, the precipitation structures at the lowest level were recognised by means of a 2D algorithm that distinguishes between convective and non-convective contribution. Secondly, the convective cells were identified using a 3D procedure quite similar to the SCIT (Storm Cell Identification and Tracking) algorithm that looks for the reflectivity cores in each radar volume. Finally, the convective cells (3D) were associated with the 2D structures (convective rainfall areas), in order to characterize the complete MCS. Once this methodology was presented the paper offers a proposal for classifying the precipitation systems, and particularly the MCS. 57 MCS structures were classified: 49% of them were identified as linearly well-organised systems, called TS (39%), LS (18%) and NS (43%). In addition to the classification, the following items were analysed for each MCS found: duration, season, time of day, area affected and direction of movement, and main radar parameters related with convection. The average features of those MCS show an area of about 25000 km², Z_max values of 47 dBz, an echotop of 12 km, the maximum frequency at 12 UTC and early afternoon and a displacement towards E-NE. The study was completed by analysing the field at surface, the presence of a mesoscale low near the system and the quasi-stationary features of three cases related with heavy rainfalls. Maximum rainfall (more then 200 mm in 6 h) was related with the presence of a cyclone in combination with the production of a convective train effect.