Search Results

You are looking at 1 - 4 of 4 items for :

  • Author or Editor: Dong Zheng x
  • Journal of Atmospheric and Oceanic Technology x
  • All content x
Clear All Modify Search
Kanghui Zhou, Yongguang Zheng, Wansheng Dong, and Tingbo Wang

Abstract

Precise and timely lightning nowcasting is still a great challenge for meteorologists. In this study, a new semantic segmentation deep learning network for cloud-to-ground (CG) lightning nowcasting, named LightningNet, has been developed. This network is based on multisource observation data, including data from a geostationary meteorological satellite, Doppler weather radar network, and CG lightning location system. LightningNet, with an encoder–decoder architecture, consists of 20 three-dimensional convolutional layers, pooling and upsampling layers, normalization layers, and a softmax classifier. The central–eastern and southern China was selected as the study area, with considerations given to the topography and spatial coverage of the weather radar and lightning observation networks. Brightness temperatures (T B) of six infrared bands from the Himawari-8 satellite, composite reflectivity mosaic, and CG lightning densities were used as the predictors because of their close relationships with lightning activity. The multisource data were first interpolated into a uniform spatial/temporal resolution of 0.05° × 0.05°/10 min, and then training and test datasets were constructed, respectively. LightningNet was trained to extract the features of lightning initiation, development, and dissipation. The evaluation results demonstrated that LightningNet was able to achieve good performance of 0–1-h lightning nowcasts using the multisource data. The probability of detection, the false alarm ratio, the area under relative operating characteristic curve, and the threat score (TS) of LightningNet with all three types of data reached 0.633, 0.386, 0.931, and 0.453, respectively. Because geostationary meteorological satellite and radar both possess the capability of capturing lightning initiation (LI) features, LightningNet also showed good performance in LI nowcasting. When all three types of data were used, more than 50% LI was predicted accurately and the TS exceeded 0.36. LightningNet’s nowcast performance using triple-source data was clearly superior to that using only single-source or dual-source data, and these findings indicate that LightningNet has good capability of combining multisource data effectively to produce more reliable lightning nowcasts.

Restricted access
Ruiyang Ma, Dong Zheng, Yijun Zhang, Wen Yao, Wenjuan Zhang, and Deqing Cuomu

Abstract

Herein, we compared data on the spatiotemporal distribution of lightning activity obtained from the World Wide Lightning Location Network (WWLLN) with that from the Lightning Imaging Sensor (LIS). The WWLLN and LIS both suggest intense lightning activity over the central and southeastern Tibetan Plateau (TP) during May–September. Meanwhile, the WWLLN indicates relatively weak lightning activity over the northeastern TP, where the LIS suggests very intense lightning activity, and it also indicates a high-density lightning center over the southwestern TP that is not suggested by the LIS. Furthermore, the WWLLN lightning peaks in August in terms of monthly variation and in late August in terms of 10-day variation, unlike the corresponding LIS lightning peaks of July and late June, respectively. Other observation data were also introduced into the comparison. The blackbody temperature (TBB) data from the Fengyun-2E geostationary satellite (as a proxy of deep convection) and thunderstorm-day data support the spatial distribution of the WWLLN lightning more. Meanwhile, for seasonal variation, the TBB data are more analogous to the LIS data, whereas the cloud-to-ground (CG) lightning data from a local CG lightning location system are closer to the WWLLN data. It is speculated that the different WWLLN and LIS observation modes may cause their data to represent different dominant types of lightning, thereby leading to differences in the spatiotemporal distributions of their data. The results may further imply that there exist regional differences and seasonal variations in the electrical properties of thunderstorms over the TP.

Restricted access
Penglei Fan, Dong Zheng, Yijun Zhang, Shanqiang Gu, Wenjuan Zhang, Wen Yao, Biwu Yan, and Yongbin Xu

Abstract

A systematic evaluation of the performance of the World Wide Lightning Location Network (WWLLN) over the Tibetan Plateau is conducted using data from the Cloud-to-Ground Lightning Location System (CGLLS) developed by the State Grid Corporation of China for 2013–15 and lightning data from the satellite-based Tropical Rainfall Measuring Mission (TRMM) Lightning Imaging Sensor (LIS) for 2014–15. The average spatial location separation magnitudes in the midsouthern Tibetan Plateau (MSTP) region between matched WWLLN and CGLLS strokes and over the whole Tibetan Plateau between matched WWLLN and LIS flashes were 9.97 and 10.93 km, respectively. The detection efficiency (DE) of the WWLLN rose markedly with increasing stroke peak current, and the mean stroke peak currents of positive and negative cloud-to-ground (CG) lightning detected by the WWLLN in the MSTP region were 62.43 and −56.74 kA, respectively. The duration, area, and radiance of the LIS flashes that were also detected by the WWLLN were 1.27, 2.65, and 4.38 times those not detected by the WWLLN. The DE of the WWLLN in the MSTP region was 9.37% for CG lightning and 2.58% for total lightning. Over the Tibetan Plateau, the DE of the WWLLN for total lightning was 2.03%. In the MSTP region, the CG flash data made up 71.98% of all WWLLN flash data. Based on the abovementioned results, the ratio of intracloud (IC) lightning to CG lightning in the MSTP region was estimated to be 4.05.

Full access
Luwen Chen, Yijun Zhang, Weitao Lu, Dong Zheng, Yang Zhang, Shaodong Chen, and Zhihui Huang

Abstract

Performance evaluation for the lightning location system (LLS) of the power grid in Guangdong Province, China, was conducted based on observation data of the triggered lightning flashes obtained in Conghua, Guangzhou, during 2007–11 and natural lightning flashes to tall structures obtained in Guangzhou during 2009–11. The results show that the flash detection efficiency and stroke detection efficiency were about 94% (58/62) and 60% (97/162), respectively. The arithmetic mean and median values for location error were estimated to be about 710 and 489 m, respectively, when more than two reporting sensors were involved in the location retrieval (based on 87 samples). After eliminating one obviously abnormal sample, the absolute percentage errors of peak current estimation were within 0.4%–42%, with arithmetic mean and median values of about 16.3% and 19.1%, respectively (based on 21 samples).

Full access