大气探测与雷电防护
Atmospheric Sounding and Lightning Protection

大气探测和雷电研究进展

1 雷电观测试验和物理过程研究

1.1 雷电野外观测试验

2012年5月中旬至9月底，大气探测研究所在广州野外雷电试验基地开展了第7年的广东闪电综合观测试验（GCOELD）。该试验在人工引雷试验场、从化市气象局和广东省气象局3个观测点实施。人工引雷试验场主要围绕人工引雷开展闪电光电磁综合测量，通信基站、风力发电机防护试验，高压输电线感应过电压特征观测以及闪电化学效应的测量，同时兼顾对自然闪电的观测；从化市气象局观测点主要对自然闪电开展综合观测，兼顾对人工引发雷电电磁辐射场的观测；广东省气象局观测点主要是针对高建筑物上雷电物理过程开展综合观测，以获取自然闪电不同类型先导（特别是上行连接先导）的物理特征。

2012年广东闪电综合观测试验在数据质量提高、观测同步、应用领域扩展等方面获得了较大进展，近距离闪电电磁场探测技术、人工引发雷电电流测量精度得到进一步发展和提高，不同站点的同步资料更加丰富，并且首次开展了闪电大气化学效应研究。

成功引发2次包含多次回击的雷电，由于改进了采集记录方法，提高了采集记录速度，改善了多站资料的匹配对应问题，2次引发雷电3个测站的电磁场观测资料收集齐全。

采用了更高性能的雷电电流采集记录装置，提高了电流数据的时间分辨率，获得的数据在信号密集区时间分辨率高达10 ns，能够展示更丰富的电流细节。同时，应用混合采样技术实现了较长时间的高精度电流信号的记录。

针对近距离闪电电磁场探测的饱和问题，改进了电磁场测量装置，提高了量程，改进后的电场慢变化测量装置量程高达±110 kV/m，电场快变化测量装置量程高达±30 kV/m。

设计实现了光辐射、宽带辐射场、低频电磁场综合观测系统，开展了自然闪电的宽带综合观测试验，获取了闪电物理过程的宽带同步观测资料，为闪电物理过程的深入分析提供了丰富的数据。

开展了闪电化学效应的观测研究，成功记录到了与闪电放电事件相伴随的大气化学成分的变化。初步的观测研究结果表明，臭氧和氮氧化物的变化与闪电事件有直接相关性，但臭氧和氮氧化物的变化具有相反的趋势。（张阳）

1.2 特殊闪电事件——袖珍云闪（CIDs）研究进展

使用宽带干涉仪首次获取了袖珍云闪（CIDs）事件的时空发展图像。对11次CIDs记录的分析表明，CIDs通道主要在垂直方向上发展。CIDs通道的垂直方向尺度在0.4～1.9 km范围。CIDs发展的视速度在(0.44～1.0)×108m/s范围，平均值为0.61×108m/s。CIDs的时空发展呈现出震荡特征，证实了之前研究工作中提出的CID通道中的电流在上下两端出现反射的猜想。通道中的电流速度在（0.56～2.6）×108m/s范围，平均速度为1.4×108m/s（图1）。

基于自行研制的重庆VLF/LF闪电定位网获得的大量2种极性的CIDs观测资料，对+CIDs和-CIDs特征的差异进行了讨论。结果表明，-CID是一种更为特殊的放电。与+CID相比，-CID会引起更大的平均电场变化，而且更远离常规闪电孤立发生。

发展了一套基于电离层反射信号的CIDs定位方法，证实-CIDs确实发生在较高的高度；同时还分析了CIDs与对流强度的关系。在9个雷暴个例中，8次雷暴中-CIDs少于+CIDs, -CIDs所占的百分比似乎随对流强度增长。尽管-CIDs相对少见,但它们出现得相当集中,大部分-CIDs都是在很短的一段时间内发生的。+CIDs也有这种特征,但没有-CIDs显著。

对广州和重庆观测到的大量CIDs事件发生高度的计算结果表明，大多数+CIDs发生高度在8～16 km，而大多数-CIDs发生高度在1619 km。很少有CIDs事件发生高度高于19 km或低于4 km。据此推断，+CIDs发生在负电荷区和上部正电荷区之间，-CIDs发生在上部正电荷区与云顶负屏蔽电荷层之间（图2）。

从2次雷暴中CIDs事件的高度变化情况来看，CIDs事件可以发生在任何2个相反极性电荷区之间。+CIDs的发生高度在-CIDs也发生的时段会相对较高。在一次雷暴的一段时间内，-CIDs的发生高度总是高于+CIDs，并且指示出+CIDs与-CIDs间的电荷区（图3）。（刘恒毅）

1.3 高建筑物上始发的上行未连接先导的特征研究

从2009年开始，在广州开展了高建筑物上雷电物理过程的观测试验。在19次闪电过程中观测到了45个上行未连接先导，这些上行先导均由闪电过程中开辟新接地点的下行梯级先导所诱发。利用高速摄像资料，统计了这些先导的一些特征，包括起始高度（范围：40～503 m；样本数：45）、起始时间（＜0.1～1.32 ms；38）、离接地点的距离（20 m～1.3 km；38）、起始时离最近下行先导头部的2D距离（99～578 m；21）、2D长度（0.48～399 m；45）和2D平均速度（（5.79～33.8）×105 m/s；22）。统计分析表明：86%（19/22）的发展速度小于1.7×105m/s；起始时间提前回击超过0.5 ms的上行未连接先导，其起始高度均超过300 m；起始高度低于300 m的上行未连接先导的长度很少超过50 m，且只会被距离在600 m以内的闪电所激发；而对于高度超过400 m的建筑物，其顶部的上行未连接先导甚至能够被1 km以外的闪电所激发；对于起始高度分别在100～200 m、200～300 m和高于400 m的上行未连接先导，诱发上行先导发生的下行先导的最远距离分别大约为350 m、450 m和600 m（图4）。（吕伟涛）

1.4 包含2次上行传播过程的触发闪电时间的光电观测

分析研究了一个产生2次正极性上行传播过程的反常人工触发闪电，第1次出现在初始阶段（即上行先导过程），第2次出现在一次负极性的下行企图先导之后。此次触发闪电没有产生回击，触发时刻，试验点上方的雷暴较弱，其内的云闪放电都产生了正向的电场变化。初始阶段的放电过程较弱，之后，直窜先导沿着由第1次上行先导建立的通道向下发展，并在距离地面约453 m的高度处停止，形成一次企图先导过程。受下行企图先导的影响，距离引流杆78 m处的地面电场在5.24 ms时间里稳定下降了6.8 kV/m，随后产生了一个新的上行通道发展（即第2次上行发展过程）。第2次上行发展过程显然是由下行企图先导触发，其在下行企图先导结束后4.1 ms起始，并沿着原有通道传播，从电流记录判断共持续了2.95 ms。在第2次上行传播过程的电流中，叠加了2个电流脉冲。第1个脉冲与该次上行先导的突然发展相联系，表现出快速的上升和下降，其峰值电流也大于第2次脉冲。第2个脉冲对应第2次上行发展的通道进入到之前下行企图先导发展的区域。可见第2次上行传播过程包含了起始的类似先导过程和随后的中和过程。该研究展现了一种新的先导触发类型，即上行放电在原有的通道中被相同通道中出现的反极性、传播方向相反的企图先导触发（图5）。（郑栋）

2 雷电业务相关研究

2.1 雷电临近预警系统的推广应用

（1）雷电监测预警预报示范应用基地的建立。以业务化推广为目的，2012年雷电临近预警系统经过进一步的技术升级，在可视化界面以及相应功能上都进行了改进和扩展，更适用于雷电预警预报的业务化应用。同时建立了北京市气象台、河北省气象台和武汉中心气象台3个雷电监测预警预报示范应用基地，开展了区域化雷电监测预警的业务平台运行试验，按照业务产品要求，实现了雷电临近预警预报服务产品的自动生成、数据共享及其网络服务功能，为雷电预警的专项服务奠定了基础。目前上述3个示范应用基地已将雷电临近预警系统应用于各自的雷电预警预报业务中，显著提高了雷电活动的临近预警准确率，尤其是在强对流天气的预警信号发布方面起到了较大的辅助决策作用。

（2）雷电临近预警系统在国家林业局森林防火预警监测信息中心业务应用。由于雷击引发的森林火灾给国家财产造成了不可估量的损失，如何有效减少雷击森林火灾的损失一直是林业部门和气象部门十分重视的头等大事。中国气象科学研究院与国家林业局森林防火预警监测信息中心合作，依据雷击森林火灾的监测、预警的具体业务需求和当前的网络基础条件，研制专门服务于林区雷击火监测以及雷击火临近预警服务方法，进行雷击森林火灾监测预警系统的开发。目前该系统在国家森林防火指挥办公室和雷击森林火灾多发的黑龙江省、内蒙古自治区，大兴安岭防火办等单位业务试运行，各单位均可通过互联网实时获取雷击火灾的监测和预警产品信息。试运行期间该系统监测到多起初发的雷击火灾，具有较好的使用价值，基本满足了雷击森林火灾监测和预警的业务需要。（姚雯）

2.2 基于触发闪电和自然闪电观测的闪电定位系统效能评估

基于2007—2011年广东从化市人工触发闪电试验和2009—2011年广州高建筑物上自然闪电观测试验获取的观测资料，对广东电网闪电定位系统效能进行了评估。

对于28次包含回击过程的触发闪电，广东电网闪电定位系统的闪电事件探测效率为89%（25/28），回击探测效率为46%（37/81）。当有2个以上探测子站参与定位计算时，平均定位误差的算术平均值和中值分别为759 m和649 m（基于33个经典触发闪电回击样本），而空中触发闪电回击定位结果与火箭发射架之间距离的算术平均值和中值分别为675 m和646 m（13个样本）。当触发闪电回击电流峰值＞15 kA 时，回击探测效率为100%（15 个样本）；当触发闪电回击电流峰值＜15 kA时，回击探测效率降为50%（7/14）；对于电流峰值＜10 kA的回击，闪电定位系统的探测效率仅为33%（1/3）。回击电流峰值直接测量结果与闪电定位系统反演结果呈现高度线性相关，相关系数达到0.92（21个样本）。

对于34次击中高建筑物的自然闪电，广东电网闪电定位系统的闪电事件探测效率为97%（33/34），对这些闪电所包含的81次回击的探测效率为74%（60/81）。当有2个以上探测子站参与定位计算时，平均定位误差的算术平均值和中值分别为633 m和453 m（54个样本）（图6）。

总体上，广东电网闪电定位系统的闪电事件探测效率为94%（58/62），回击探测效率为60%（97/162）；当有2个以上探测子站参与定位计算时，平均定位误差的算术平均值和中值分别为710 m和489 m（87个样本）；当剔除1个最大异常值时，回击电流幅值相对探测误差在0.4%～42%之间，算术平均值和中值分别为 16.3%和19.1%（21个样本）（图7）。（张义军）

2.3 1997—2009年全国雷电灾害特征统计

利用全国雷电灾害数据库资料，分析了1997—2009年我国雷电灾害在人员伤亡、财产损失等方面的统计特征。1997—2009年，我国共发生雷电灾害61614起，发生雷击人员死亡事故5033起，雷击人员受伤事故4670起。在空间分布上，我国雷电灾害发生频率呈现出东部沿海和南部地区高于西部地区的特点。在时间分布上，雷电灾害主要发生在夏季的7—9月，而冬季的10月至次年3月雷电灾害相对较少。雷电灾害的时空分布特征与我国闪电频次的时空分布相一致。雷电灾害和人员伤亡事故在1997—2007年呈现逐年上升趋势，2008年开始有所下降。我国每年每百万人口的雷击死亡率和受伤率分别为0.31和0.28。农村人口占雷击死亡和雷击受伤人数的51%和29%，农民是雷电灾害的主要受害者。雷击损失行业和雷击环境的分析表明，民用行业在雷灾事故受损行业中占有较大比重，农田是雷击伤亡事故的主要发生地（图8）。（张文娟）

3 地基云自动化观测技术和方法研究

地基全天空云观测系统的稳定性有了较大提升。引入PLC模块实现了观测系统中各种机械运动的控制和设备内部温湿度的监测，采用光纤转换模块解决了观测设备和中央控制系统之间的远距离传输问题。

重新设计开发了地基全天空云观测系统的软件，新版观测软件界面友好，功能丰富，增加了天顶方向云底高度、太阳位置信息、日出日落时间和日照时数等的计算和显示（图9）。

2012年8月起，在广东从化市气象局和西藏自治区日喀则气象局各架设了一套地基全天空云观测系统开展对比观测试验，并实现了系统的远程控制和数据传输。

研究了两种卷云检测算法，通过引入数学形态学和Markov随机场模型较大地改进了全天空图像中卷云检测的精度。在云状识别方面，研究了一种新的器测云图分类方案，先将云分成积状云、层状云和卷云，通过模式识别方法先将这3类云分开，再结合激光云高仪获取的云底高度信息，根据云底高度将云分成低云、中云和高云3族，则可得到较具体的云图分类结果。（杨俊）

图1 一次-CIDs事件VHF辐射源的二维方位角-仰角定位结果随时间的变化Fig1 Azimuth and elevation mapping of VHF radiation sources versus time for a -CIDs event

图2 广东和重庆观测到的致密云闪高度分布(广东：1318+CIDs，625-CIDs；重庆：5489+CIDs，1400-CIDs)Fig2 Distributions of discharge heights of CIDs observed in Guangzhou (a) and Chongqing (b)

图3 重庆一次雷暴过程中致密云闪高度随时间的分布特征(蓝色和绿色的三角形分别代表每15个连续的+CIDs或-CIDs的平均高度，矩形阴影区域表示没有-CIDs 发生的时间段，几乎将+CIDs和-CIDs分开的黑色曲线表示上部正电荷区的可能位置)Fig3 CID discharge heights in a thunderstorm in Chongqing. (Blue and green triangles represent average heights for each 15 successive positive CIDs and negative CIDs, respectively. Shaded rectangles indicate the periods when no negative CID is produced. Black curve represents the possible location of the upper positive charge layer, dividing almost all positive CIDs and negative CIDs into two sides)

图4 2010年6月21日广州高建筑物上一次闪电的高速摄像记录（在此个例中观测到4个上行未连接先导，UUL09～UUL12）Fig4 The high-speed images of a lightning fash striking a tall structure in Guangzhou on 21 June 2010. (Four unconnected upward leaders, UUL09—UUL12, are observed in this case)

图5 下行负极性企图先导((a)～(h))和随后的第2次上行正极性传播过程((i)～(l))((i)和(h)间隔4.1 ms）Fig5 Negative downward aborted leader ((a)-(h)) and the subsequent second positive upward propagation ((i)-(l)) (There is a 4.1-ms interval between (i) and (h))

图6 闪电定位系统对54次击中高建筑物的回击的定位偏差Fig6 The location errors of LLS reports for 54 return strokes on tall structures. The origin corresponds to the ground strike point

图7 触发闪电回击电流峰值的直接测量结果与定位系统反演结果的对比Fig7 LLS-reported vs directly measured peak currents in the triggered lightning experiment

图8 我国雷击人员伤亡事故排序Fig8 The rank of each province in the rate of lightningrelated casualties in mainland China

图9 新版地基全天空云观测系统软件界面Fig9 The interface of the new CAMS_TCI software

Progress in Atmospheric Sounding and Lightning Research

1 Lightning feld experiment and physics process research

1.1 Lightning feld experiment

Guangdong Comprehensive Observation Experiment on Lightning Discharge (GCOELD) has been conducted by the Institute of Atmospheric Sounding (IAS) at the feld experiment site for lightning research and testing in Guangdong Province from 15 May to 20 September 2012. The observations were carried out at the test site of triggered lightning, Conghua Municipal Meteorological Bureau, and Guangdong Provincical Meteorological Bureau. In the test site of triggered lightning, the main experiments include the comprehensive observation of optical radiation and electromagnetic feld, the protection of communication base station andwind generators, the observation of induction overvoltage for high voltage transmission, and the observation of lighting chemical effect. At the Meteorological Bureau of Conghua, natural lightning and triggered lightning are observed. At the Meteorological Bureau of Guangdong Province, the comprehensive observation of lightning striking on tall structures is conducted in order to obtain the physical characteristics of different leaders (especially the upward connecting leader) in natural lighting fashes.

GCOELD in 2012 has made great progress in the improvement of data quality, the synchronization of multi-station observation, and the extension of application area. The saturation problem during close electromagnetic detection and the measurement precision of current data have been improved, and the synchronous data in different observation sites have been enriched. In addition, triggered lightning as a tool has been applied to the research of atmospheric chemistry. Detailed results are shown as follows.

Two lighting fashes accompanied by many return strokes have been successfully triggered. Based on a new method of data acquisition, the speed of data acquisition has been improved. As a result, the synchronous data of three observation sites for two triggered lightning have been collected.

The time resolution of lightning current measurement has been improved by using the acquisition and recording equipment with higher performance. As a result, the time resolution is 10 ns for dense signals and more detailed information is obtained. At the same time, longer recording time can be achieved by application of a hybrid sampling method.

The detection system of electric feld range has been improved in order to resolve the saturation problem for close electromagnetic feld detection. The measurement range of slow electric feld change is ±110 kV/m, and the corresponding value is ±30 kV/m for the improved system of fast electric feld change.

The detection system of optical radiation, broadband radiation, and low frequency electromagnetic feld has been developed. During the broadband comprehensive observation, the synchronous data of lightning sunphysical process have been collected, which will beneft the further analysis of lightning process.

The observation of lighting chemical effect has been conducted. The chemical composition in ambient atmosphere varies when lightning flashes occur. The preliminary results show that there is a correlation between the concentration change of ozone and nitrogen oxide and the lightning event, but the change tendency of ozone concentration is opposite to that of nitrogen oxide. (Zhang Yang)

1.2 Progress in the research of compact intracloud discharges (CID)

The CID channel evolution images obtained by using VHF broadband interferometers are presented for the frst time. Analysis of 11 CIDs shows that the channels of CIDs develop mainly in the vertical direction. The vertical scale of CIDs is in the range of 0.4-1.9 km. The average apparent speed of CIDs is in the range of (0.44-1.0)×108m/s with a mean value of 0.61×108m/s. The temporal-spatial evolution of the radiation sources of the CID shows an oscillation pattern, confrming the previous prediction that there is an oscillating current being refected at the two ends of the CID channel. The estimated speed of the current wave in the CID channel is in the range of (0.56-2.6)×108m/s with a mean value of 1.4×108m/s (Fig1).

On the basis of a large number of CIDs of both polarities recorded by our VLF/LF lightning location network, characteristic differences between +CIDs and −CIDs are discussed. The results reveal that −CID is a more special type of discharge. Compared with +CIDs, −CIDs produce larger electric feld changes on average, and they are more isolated from other discharge processes.

A locating method based on ionospheric refection pairs of CIDs is developed, which confrms that−CIDs do occur at higher altitudes. The relationship between CIDs and convective strength is also analyzed. Out of nine storms analyzed in this study, eight produce fewer −CIDs than +CIDs. The percentage of −CIDs seems to increase with the convective strength. Although −CIDs are relatively rare, their occurrences are more temporally compact, that is, a large portion of −CIDs are produced in a very short period. +CIDs also have this characteristic, but not as pronounced as that of −CIDs.

Discharge heights of thousands of CIDs observed in Guangzhou and Chongqing of China are calculated. The result shows that most positive CIDs occur between 8 and 16 km while most negative CIDs occur between16 and 19 km. Very few negative CIDs are above 19 km or below 4 km. It is inferred that positive CIDs are produced between main negative charge layer and upper positive charge layer while negative CIDs are produced between upper positive charge layer and negative screening charge layer at the cloud top (Fig2).

Variations of CID discharge heights in two thunderstorms are analyzed. It seems that CIDs can be produced at any position between corresponding charge layers. Positive CIDs are generally higher in the periods when negative CIDs are also occurring. For a given short time period during a single thunderstorm, negative CIDs are always observed to occur at higher altitudes than positive CIDs, indicating a dividing charge layer between positive CIDs and negative CIDs. We believe CIDs are produced in vigorous convective surges that develop to the height comparable to the discharge height of CIDs (Fig3). (Liu Hengyi)

1.3 Characteristics of unconnected upward leaders initiated from tall structures

To study the processes of lightning flashes striking on tall structures, a field experiment has been conducted in Guangzhou since 2009. 45 unconnected upward leaders (UULs) that occurred in 19 downward negative fashes are analyzed. Each observed UUL is initiated by a downward stepped leader before a new strike point is struck. For each UUL, several parameters are determined by using highspeed images∶ inception height, inception time prior to return stroke (RS), horizontal distance from the fash's strike point, two-dimensional (2D) distance between the nearest downward leader branch tip and the UUL’s inception point at its inception time, 2D length, and 2D average propagation velocity. Their corresponding values range from 40 to 503 m (number of samples∶ 45), ＜0.1 to 1.32 ms (38), 20 m to 1.3 km (38), 99 to 578 m (21), 0.48 to 399 m (45), and 5.79×105to 33.8×105m/s (22), respectively. Statistical analysis shows that 86% (19/22) of the velocities are smaller than 1.7×105m/s; no UUL with an inception time prior to RS greater than 0.5 ms is initiated from a structure lower than 300 m; those UULs with inception heights lower than 300 m seldom exhibit lengths longer than 50 m and can only be initiated by fashes within approximately 600 m, while those higher than 400 m can even reach several hundred meters and be initiated by fashes over 1 km away. The maximum distances for the downward leaders to attract the UULs with inception heights from 100 to 200 m, 200 to 300 m, and over 400 m are approximately 350 m, 450 m, and 600 m, respectively (Fig4). (Lü Weitao)

1.4 Optical and electrical observations of an abnormal triggered lightning event with two upward propagations

This study investigates an abnormal artificially triggered lightning that produced two times positive upward propagations∶ one during the initial stage (i.e., the upward leader (UL)) and the other after a negative downward aborted leader (DAL). The triggered lightning was performed in a weak thunderstorm over the experiment site and did not produce any return stroke process. All the intracloud lightning around the experiment site produced positive electric field changes. The initial stage was of weak discharge process. After that, a downward dart leader propagated along the original channel produced by the 1st UL and ended at a height of about 453 m, forming a DAL. Infuenced by the DAL, the electric feld at a point located 78 m from the rod showed a steady reduction by about 6.8 kV/m over 5.24 ms before the initiation of a new upward channel (i.e., the 2nd upward propagation (UP)). The 2nd UP, which started at about 4.1 ms after termination of the DAL and propagated along the original channel, was triggered by the DAL and sustained for about 2.95 ms according to the current record. Two current pulses were superimposed on the current of the 2nd UP. The frst pulse, related to the sudden initiation of the 2nd UP, showed a more rapid increase and decrease, and large peak value, than did the second pulse, related to the development of the 2nd UP into the area where the DAL had propagated. The 2nd UP contained the similar-to-leader process and the following neutralization process. This study represents a new type of triggering leader, in which a new upward discharge is triggered in an established channel by the aborted leader propagating along the same channel with opposite polarity and propagation direction (Fig5). (Zheng Dong)

2 Research related to lightning operational work

2.1 Promotion and application of the Lightning Nowcasting and Warning System (LNWS)

(1) Establishment of demonstration bases for lightning monitoring, warning, and forecasting. For the purpose of operational application, the LNWS conducted further technology upgrades in 2012. With improvement and expansion of visual interface, it is more suitable for the operational lightning warning forecast. Meanwhile, three demonstration bases for lightning monitoring, warning, and forecasting have been set up in the Beijing Meteorological Center, Hebei Meteorological Center, and Wuhan Meteorological Center, and running tests of these regional operational platforms have been carried out. In accordance with the requirements of operational products, the system automatically generates service products for lightning warning and forecasting and has the function of data sharing and network services. Presently, the above three demonstration bases have applied the LNWS to their own lightning warning operation and improved the nowcasting accuracy, especially in the early warning of severe convective weather.

(2) Operational application of the LNWS in Forest Fire Warning at the Monitoring Information Center of the State Forestry Administration. Considering the grave losses to state property by the lightning-ignited forest fres every year, how to effectively reduce the losses of forest fres caused by lightning has been a top priority concern for both forestry and meteorological administrations. Based on the specifc operational needs and current network infrastructure, the Chinese Academy of Meteorological Sciences in cooperation with the Forest Fire Warning and Monitoring Information Center of the State Forestry Administration, established a specialized service for monitoring and nowcasting lightning-ignited forest fres, and developed a Lightning-Ignited Forest Fire Monitoring and Warning System. The system has currently been put into operational use in the State Forest Fire Prevention Offce in Hei Longjiang Province, Inner Mongolia Autonomous Region, and the Daxing'anling Mountain Range. All the departments could obtain the lightning-ignited fre monitoring and warning products in real time via the Internet. During the test run period, the system captured several initial lightning-ignited fires, which demonstrates the good value of this system in monitoring and nowcasting of lightning-caused forest fres. (Yao Wen)

2.2 Performance evaluation for a lightning location system based on observations of artifcially triggered

lightning and natural lightning fashes

Performance evaluation for the lightning location system (LLS) of the power grid in Guangdong Province of China was conducted based on observational data of the triggered lightning fashes obtained in Conghua of Guangdong during 2007—2011 and natural lightning fashes over tall structures obtained in Guangzhou during 2009—2011.

For 28 artificially triggered lightning flashes (each of them contains one or more return strokes), the LLS’s flash detection efficiency and stroke detection efficiency are approximately 89% (25/28) and 46% (37/81), respectively. The location accuracy of LLS reports is analyzed based on those location retrieval results using more than two sensors. For 33 return strokes occurring in classical triggered lightning fashes, the arithmetic mean and median values of the LLS’s location error are estimated to be approximately 759 and 649 m, respectively; and for 13 return strokes occurring in altitude triggered lightning fashes, the arithmetic mean and median values of the distances between the LLS’s records and the rocket launcher are approximately 675 and 646 m, respectively. When the absolute peak current of a return stroke was greater than 15 kA, the detection effciency was 100% (15/15), but it decreased to 50% (7/14) when the peak current was less than 15 kA, and only 33% (1/3) in cases where the peak current was less than 10 kA. There is a strong positive linear relationship between the directly and LLS estimated and directly measured peak currents, with the correlation coeffcient being 0.92 (for 21 samples).

For 34 natural lightning fashes over tall structures in Guangzhou, the LLS’s fash detection effciency is approximately 97% (33/34). For the 81 return strokes included in these fashes, the LLS’s stroke detectionefficiency is approximately 74% (60/81). If more than two reporting sensors are involved in the location retrieval, the arithmetic mean and median values for location error are estimated to be approximately 633 and 453 m, respectively (for 54 samples) (Fig6).

Totally, the LLS’s fash detection effciency and stroke detection effciency are approximately 94% (58/62) and 60% (97/162), respectively. The arithmetic mean and median values for location error are estimated to be approximately 710 and 489 m, respectively, when more than two reporting sensors are involved in the location retrieval (for 87 samples). After eliminating one obviously abnormal sample, the absolute percentage errors of peak current estimation are within 0.4%-42%, with arithmetic mean and median values of approximately 16.3% and 19.1%, respectively (for 21 samples) (Fig7). (Zhang Yijun)

2.3 Lightning casualties and damages in China from 1997 to 2009

The lightning-related fatalities, injuries, and property damages reported in China from 1997 to 2009 are summarized by using the National Lightning Hazards Database. The characteristics of the incidents including 5033 deaths, 4670 injuries, and 61614 damage reports are analyzed. For the spatial distribution of lightning disasters in China, the eastern costal and southern areas have more frequent lightning disasters than the western areas. Lightning disasters mainly occur in summer months from July to September while fewer damages occur in winter months from October to March, which correlate signifcantly with the temporal variability of lightning frequency in China. Lightning-related casualties and damages in China have increased for the period of 1997—2007, and then began to decrease since 2008. The national fatalities and injuries per million people per year are 0.31 and 0.28, respectively. Rural people accounts for 51% and 29% of all lightning fatalities and injuries, which makes residents in agricultural and rural area the major lightning victims. Characteristics of lightning disasters and correlative factors are also studied, including hazard affected industries and locations. The results show that civil industry has the worst property loss and farmland is the largest category in lightning-caused casualty locations (Fig8). (Zhang Wenjuan)

3 Progress in research on ground-based cloud automatic observation

The stability of the ground-based Total-sky Cloud Imager (CAMS_TCI) has been greatly improved after accomplishing the monitor and control of mechanical motion and internal temperature and humidity by using PLC module. The problem of long-distance transmission between the observation equipment and the central control system has been solved by adopting fber optic conversion module.

The software of the CAMS_TCI has been re-designed. The new observation software interface is friendly and feature-rich, with added information such as the height of the cloud base, the position of the sun, sunrise and sunset time, sunshine, and so on.

Since August 2012, two sets of CAMS_TCI have been installed at the Conghua Meteorological Bureau and Shigatse Meteorological Bureau to carry out a comparative observation experiment, respectively. A remote control and data transmission technique between the client and the host were used in the experiments.

By introducing mathematical morphology and Markov random field model, two cirrus detection algorithms have been proposed, which have improved the cirrus cloud detection accuracy of the total-sky image greatly. In the cloud type recognition, a new instrumental classifcation scheme has been put forward, which classifies the cloud into cumuliform cloud, stratiform cloud, and cirrus cloud. The three types of clouds were identifed through pattern recognition, then combined with the cloud base height from the laser ceilometer, which could divide the cloud into low cloud, middle cloud, and high cloud. Specifc cloud type classifcation results could thus be obtained (Fig9). (Yang Jun)