To get more insights into the charge structure inside thunderstorms over high plateau regions, this poses significant challenges. Based on the data from the high-precision lightning VHF interferometer, C-band Doppler radar and multi-band electromagnetic field observation, this study presented the complete evolutionary characteristics of charge structure in an atypical bottom-heavy thunderstorm producing solely intra-cloud (IC) lightning over the Tibetan Plateau (TP). A persistent inverted dipole and a tripolar charge structure with a larger-than-usual lower positive charge center (LPCC) coexisted within this thunderstorm. At the initial and the sustained weak convection stages, the thunderstorm demonstrated a negative dipolar charge structure. As the convection further developed into a mature stage, an additional positive charge region appeared in the upper level, forming a tripolar charge structure with a larger LPCC at the bottom of an intense convective region. In contrast, a neighboring convective region preferentially developed the upper-level dipole charge structure rather than forming an LPCC. The IC flashes, with a maximum of 4 fl/min, predominantly occurred as negative IC flashes originating from the lower negative dipole, while few positive IC flashes occurred between the upper dipole charge structure. The LPCC was primarily associated with positive charged graupel particles. The middle negative charge region involved substantial negative charge carried by ice crystals and snow aggregates with graupel particles acting as the charge carrier in its upper part. The upper positive charge region was mainly contributed by ice crystals and snow aggregates.

This study utilizes lightning monitoring data from the Worldwide Lightning Location Network (WWLLN) spanning 2010 to 2022, combined with grid-based precipitation data from the CN05.1 network of the Chinese Academy of Sciences and reanalysis data (European Centre for Medium-Range Weather Forecasts Reanalysis 5th Generation, ERA5) from the European Centre for Medium-Range Weather Forecasts (ECMWF), to investigate lightning activity in the Guangdong-Hainan region of China, an area with the highest frequency of lightning strikes. By analyzing the spatiotemporal characteristics of lightning activity, the study explores the relationship between lightning and various atmospheric environmental factors. The results indicate that lightning exhibits distinct seasonal (more frequent in summer than winter), regional (more frequent in the south than the north), and geographical (more frequent along coastlines than inland) patterns. Lightning activity is strongly correlated with Convective Available Potential Energy (CAPE), precipitation (P), the product of CAPE and precipitation (CAPE × P), and the K index. The spatial distribution of lightning is most strongly correlated with atmospheric elements in western Guangdong. Further analysis reveals that lightning is most likely to occur when Convective Inhibition Energy (CIN) ranges from 50 to 100 J kg⁻¹ and CAPE ranges from 500 to 1500 J kg⁻¹. Both excessively low CIN values and excessively high CAPE values result in reduced lightning activity. Lightning is most active under relative humidity (RH) conditions between 60% and 90%, with very low lightning formation occurring when RH falls below 30%. In Hainan, the probability of lightning occurrence is higher than in other regions under similar RH conditions or high CIN values.

Extreme precipitation and lightning strikes exert substantial economic losses and human casualties. Under climate change, the frequency, intensity, and duration of these extreme weather events are projected to change significantly. Based on data from 345 meteorological stations across China from 1979 to 2013, this study examines the relationship between extreme precipitation and thunderstorm days. In China, thunderstorm days account for over 50

Abstract Based on combined VHF interferometer and fast/slow antenna measurements, we document an abnormal bolt‐from‐the‐blue lightning flash, which included two strokes. There are four successive attempted dart leaders following the first return stroke. The last detected downward leader stopped advancing at approximately 94 m and decayed. Then, a fast upward breakdown initiated 2.29 ms after it stopped, developing along the preexisting channel toward the cloud, propagating with a speed of 4.9 × 107 m/s along the vertical portion of the channel. Electrostatic field changes revealed that negative charge transferred by previous attempted dart leaders had accumulated within the decaying channel. The accumulated charge was comparable to that later transferred to ground by the delayed return stroke, which itself was accompanied by a long continuing current. We term this event ‘delayed return stroke’. This phenomenon illustrates the independent influence of residual charge within the decaying channel on later discharge processes.

Lightning constitutes a significant threat to human life and property. Effective lightning forecasting significantly mitigates associated losses. However, a critical gap persists in the lack of methods capable of providing precise lightning forecasts at high spatial resolution (1 & times;1 km). To address this limitation, we propose high-resolution U-Net (HRUnet), a novel deep learning (DL) model integrating the U-Net encoder-decoder architecture with inceptionres-convolutional block attention module (IR-CBAM) block for temporal feature enhancement and PixelShuffle for spatial resolution preservation. Furthermore, we introduce a specialized focus loss function to alleviate the severe class imbalance inherent in lightning prediction tasks. HRUnet was evaluated using weather radar and lightning data from Binzhou, Shandong Province, China. Experimental results demonstrate that: 1) the IR-CBAM module, PixelShuffle module, and focus loss function yield improvements of 7.22%, 0.26%, and 4.02%, respectively, in average equitable threat score (ETS) compared to the baseline U-Net; collectively, HRUnet achieves a 10.86% absolute ETS improvement and 2) on the test set, HRUnet attains performance metrics of 0.9843 hit rate (HR), 0.6015 probability of detection (POD), 0.2856 false alarm rate (FAR), and 0.4713 ETS. HRUnet outperforms the other four state-of-the-art DL models and demonstrates the best capability in forecasting the initiation, development, and dissipation of lightning clusters among them.

Affected by solar radiation in space, the FY-4A Lightning Mapping Imager (LMI) detection array exhibits daily periodic thermal expansion and contraction, leading to deviations in lightning positioning accuracy. While LMI’s detection efficiency is higher at night, the dual edge matching algorithm, which relies on surface features for correction, does not perform well during nighttime (around 3 pixels). Analysis shows that most of the lightning data corrected by this method exhibit significant deviations from the actual lightning locations in practical applications. Therefore, this paper proposes a new correction method based on high precision ground-based lightning location data from the 2019 summer World Wide Lightning Location Network (WWLLN) and the Beijing Broadband Lightning Network (BLNET). Using these datasets as reference standards, the periodic deviation of LMI is determined, and a correction curve is derived using a weighted Gaussian fitting approach. This method further improves the nighttime lightning location accuracy of LMI on the basis of the current operational algorithm. The results demonstrate that the corrected LMI data significantly reduces the positioning errors, with an accuracy within ±1 pixel in the Beijing area, as an example.

A novel high-speed area-array imaging instrument, the Dynamic Grayscale-Triggered Hydrometeor Particle Observation System (DGHOS), is presented for comprehensive observation of hydrometeor particles. DGHOS is capable of directly imaging the descent trajectories of hydrometeor particles while simultaneously capturing both their kinematic and morphological characteristics. To minimize sensitivity to variations in particle distance from the lens, the system employs a high-resolution area-array camera integrated with a bi-telecentric optical system. A frame rate of up to 400 fps is achieved by reducing the ROI (region of interest). The use of a high-intensity monochromatic LED area light source effectively avoids chromatic dispersion. Through a grayscale-change-triggered acquisition mode, the volume of stored data is reduced by approximately 98.5% in the presented rainfall case compared with continuous recording at the same frame rate. Multiple parameters, including position, velocity, equivalent diameter, and axis ratio, are extracted using Canny edge detection and centroid-matching algorithms. System performance was evaluated through calibration experiments and natural hydrometeor particle observations, yielding a spatial resolution of 11.3 lp mm−1 and a scale factor of approximately 39 μm px−1. Under natural conditions, extensive populations of raindrops and snowflakes were successfully captured by DGHOS. Representative images of both types are presented, accompanied by a detailed quantitative analysis focused on their dynamics.

In recent years, high-impact thundersnow events have occurred frequently across China. To deepen the understanding of thundersnow mechanisms and enhance forecasting capabilities, this study conducts a detailed investigation of a thundersnow event on December 7, 2024, over the Shandong Peninsula using radar, lightning, atmospheric electric field data, and ERA5 reanalysis. Key findings include the following aspects: (1) This thundersnow event occurred under typical sea-effect (cold-air advection) snowfall conditions over the Bohai Sea, with a temperature difference of 21 °C between the 850 hPa level and the sea surface. The convective instability below 850 hPa, forced by positive vorticity advection and cold advection in the mid-upper levels, generated strong and widespread upward motion below 600 hPa. The lifting of warm-moist air near the sea surface released unstable energy, resulting in thunderstorms. (2) Lightning flashes during this sea-effect snowfall were relatively sparse. Their frequency and polarity distribution resembled those of weak summer thunderstorms. Cloud-to-ground lightning flashes preceded or coincided with heavy snowfall episodes and occurred near or upstream of areas of heavy snowfall. (3) Radar echoes exhibited distinct convective characteristics, with the maximum reflectivity reaching 48.5 dBZ and the echo tops height exceeding 5.2 km. The snow echoes were primarily distributed in a band parallel to the coastline, and the "train effect" was the main cause of heavy snowfall. (4) Polarimetric radar observations indicate abundant pristine ice crystals, graupel, and supercooled water within the convective cells, while ERA5 provides environmental thermodynamic and dynamic conditions consistent with their development. The cloud system possessed the necessary microphysical and dynamic conditions for non—inductive electrification.

Comparative analysis of positive and negative altitude-triggered lightning striking a 30-m tower beneath 35-m floating wires reveals detailed bidirectional leader processes. It provides the first observational evidence of bidirectional leader behavior in positive altitude-triggered lightning and highlights the connection process's driving effect on leader initiation/development. In positive triggered lightning, the bidirectional leader system first emerged from the wire bottom with positive polarity, rapidly connected with the tower and generated an impulsive current wave initiating an upward negative leader from the wire top. The wire ends generated different E-field distortions, but less significant than the different initiation thresholds of positive and negative leaders. Consequently, the positive leader always initiates first for both triggering cases, regardless of background E-field direction and formation position. Wire end's E-field distortion alone was insufficient to initiate negative leader. Besides slow charge accumulation from the opposite end's positive leader, impulsive current surge is another effective/indispensable abrupt driver.

The Qinghai-Tibetan Plateau is the highest in the world, and it is located in an area jointly influenced by the Indian monsoon and westerly winds. With its unique thermal and dynamic effects and complex topographical features, thunderstorms and convective activities occur frequently in the summer over most of the Plateau. The thunderstorm days in the central and eastern Plateau are similar to that of the southeastern coastal region with the most thunderstorm processes and lightning activities in China. However, the life cycle of thunderstorms over the Plateau is usually shorter, and their convective intensity is weaker, compared with those over prominently lower altitude regions. Thunderstorms over the Plateau not only show unique convective structures, but their electrical characteristics also differ significantly from those of thunderstorms over plain areas. Due to the difficulty of observation in the Plateau regions and the limitations in lightning detection technologies, it remains a challenge to understand the intercloud charge structure and lightning initiation of the thunderstorms over the main body of the Plateau region. By utilizing very high frequency (VHF) broadband interferometer lightning mapping technique, weather radar, and radio sounding, a thunderstorm process that produced two rare "bold- from-the-blue" flashes was observed in the Lhasa area located in the central Plateau. The mechanism of the "bold-from-the- blue" flashes was revealed through accurate mapping of the discharge processes in high spatial and temporal resolution, and the corresponding charge structure characteristics inside the thunderstorm were also retrieved. At the initial stage of the thunderstorm, it exhibited a negative dipole charge structure with an upper negative and bottom positive pattern, which is completely different from the evolution of charge structure in the thunderstorms over prominently lower altitude regions. When the thunderstorm developed into its mature stage, it exhibited a tripolar charge structure stacked with positive, negative, and positive charge regions from the top to bottom of the thunderstorm. These regions corresponded to environmental temperature zones below -30 degrees C, from -30 degrees C to -15 degrees C, and above -10 degrees C, respectively. The "bold-from-the- blue" flashes were associated with the upper dipole. The discharge originally initiated between the central negative and the upper positive charge region, with the positive leader developing downward to the lower negative charge region and the negative leader upward to the upper positive charge region. When the upward negative leader reached the upper positive charge region, it propagated horizontally and got out from the body of the small storm cell. The discharge ultimately became the "bold-from-the-blue" discharge process, spreading away from the cloud body into the clear sky area and striking the ground. The grounding points of the two "bold-from-the-blue" flashes were 3.6 and 3.8 km from the precipitation edge of the storm, respectively. The horizontal scale of the thunderstorm cell was relatively small with positive charge regions in both the upper and lower parts of the cloud. The unbalanced upper dipole in the smaller storm body, where the upper positive charge region was weaker than the lower negative charge region, was the main reason for triggering the "bolt-from-the-blue" flashes. This study not only sheds light on the mechanism of the rare "bold-from-the- blue" discharge process, but also provides direct observational evidence on the diversity of intercloud charge structure and lightning activity characteristics inside the thunderstorm over high-altitude regions.

Accurate lightning forecasting plays a crucial role in mitigating the hazards posed by severe convective weather. This study presents an AutoWeight-Net for radar-based lightning prediction, leveraging S-band radar data from Binzhou, Shandong Province, collected between June 1 and August 31, 2024. The radar parameters used include CR, VIL, PPI (0.5° and 1.5°), and TOPS (18 dBZ and 45 dBZ thresholds). The AutoWeight-Net is a U-Net based model which has the following innovation: (1) Adding a channel-wise weighting module at the model’s input, enabling the network to learn the importance of each radar feature. (2) We replace the convolution layers in the lowest resolution of the U-Net with attention layers, allowing the model to focus on areas with thick cloud cover, which are more susceptible to lightning. (3) A custom loss function is designed to address the severe class imbalance between lightning and non-lightning samples. The model predicts the probability of lightning occurrence within the next 90 minutes, at 30-minute intervals, and at a spatial resolution of 1 km × 1 km. Experimental results demonstrate 99% accuracy, 62.8% precision, and 73.5% false detection rate, underscoring the effectiveness of the proposed improvements in enhancing lightning forecast performance. These findings hold promise for real-time operational forecasting and risk management in regions prone to convective weather.

Vertical transport carries airmasses from different sources of tropospheric O 3 including photochemical production and stratospheric intrusions, and is crucial for vertical O 3 variability. Based on temporally dense ozonesondes over northern China, this study reports an anomalous vertical O 3 distribution characterized by a “bottom‐heavy” structure. Specifically, O 3 well exceeded the normal values in the middle‐to‐lower troposphere due to the stratospherically intruded airmasses (SIA), but was sharply reduced in the upper troposphere and lower stratosphere region because of convectively lofted air transported from the Tibetan Plateau. Guided by the discovery of such a vertical O 3 structure, we further assess the SIA contribution to tropospheric O 3 using multi‐year AIRS satellite observations, and find that SIA appear frequently during summer and lead to short‐term O 3 enhancements 35% above the normal values at 500 hPa. These results reveal the linkages between vertical O 3 variations and synoptic processes, highlighting the non‐negligible contribution of SIA to tropospheric O 3 .

The inverted tripole charge structure in thunderstorm over the central Tibetan Plateau was discovered for the first time, primarily through observations from lightning very high frequency interferometer capable of high‐precision lightning channel mapping. The dominant cell exhibited an inverted tripole charge structure initially, characterized by a negative charge region at temperatures near 0°C, a main positive charge region between −30°C and −5°C, and an upper negative charge region at T < −20°C. The cell's rear portion exhibited a normal tripole before detaching, leaving a pure inverted tripole. Dissipation of the lower negative charge transitioned the structure to an inverted dipole, consisting of an upper negative ( T < −20°C) and lower positive ( T > −20°C). Throughout this thunderstorm, no positive cloud‐to‐ground (+CG) flashes were detected, while five −CG flashes were recorded. Among 109 intracloud (IC) flashes detected, 90% occurred between the upper inverted dipole. Radar reflectivity showed that this thunderstorm was more intense than conventional plateau thunderstorms.

Quickly identifying and classifying lightning waveforms is the foundation of lightning forecasting and early warning. In this paper, based on the electric field observation of the Beijing lightning location website of the Institute of Atmospheric Physics, Chinese Academy of Sciences, a recognition and classification method of pulse signal waveform based on Convolutional Neural Network(CNN) algorithm is designed and implemented. The CNN network model and its parameters were optimized from three aspects: dataset, model parameters, and network structure, achieving a recognition rate of over 90%. The effects of various optimization terms and their different optimization orders on the training time of the model were studied. The results indicate that the CNN algorithm is suitable for the classification and recognition of lightning electric field (LEF) waveforms. Optimization can significantly improve recognition rate. The optimization method of fitting idealized waveforms can reduce noise in the dataset and significantly improve recognition rate, indicating that noise has a significant impact on waveform recognition. Therefore, it is necessary to perform noise preprocessing before recognition. The optimization has a huge impact on training efficiency, increasing training time by about 51% after optimization, but the influence of optimization order on it can be ignored.

The main charge region in thunderstorms over Lhasa city with an elevation of 3700 m is investigated by using a VHF interferometer, incorporating with fast antenna, weather radar and cloud-to-ground lightning location. The evolution of charge structure and its effects on lightning discharges were discussed in a bottom-heavy thunderstorm. During the early developing stage, the thunderstorm exhibited an inverted dipolar charge structure with negative charge center over the positive, and lower negative intracloud (IC) lightning occurred in between. Then an upper positive charge region appeared as the convection intensifying, and the charge structure exhibited obvious tripolar pattern and with large lower positive charge center (LPCC), and fewer positive IC discharges occurred in the upper dipole but lower negative IC lightning still dominated. As the thunderstorm entered the later mature stage, both negative IC between the lower dipole and positive IC between the upper dipole observed simultaneously. With gradually depleting of the positive charge carriers by precipitation, the LPCC weakened, the positive IC lightning between the upper dipole dominated, and two negative CG flashes were able to occur. In the later stage, positive IC dominated, although not much. The study further confirms the previous conclusion (Qie et al., GRL, 2005) that weak thunderstorms are characterized by a bottom-heavy charge structure, and in the vigorous stage of thunderstorm, it may exhibit tripolar charge structure with a large LPCC, which has a significant impact on lightning types.

The Tibetan Plateau (TP) is one of the world's most climate‐sensitive regions. Thunderstorms represent a major type of precipitation system across the TP and contribute significantly to regional rainfall. This study generated newly a continuous thunderstorm data set clustered by applying the density‐based spatial clustering of application with noise algorithm to lightning data obtained from the World Wide Lightning Location Network from 2010 to 2024. On the basis of this data set, a significant increase in thunderstorm activity was found over the TP in recent two decades, with the maximum rate of increase reaching 2.7 10 −4 km −2 thunderstorms per year in the south‐central TP. At the same time, a weak decreasing trend of the thunderstorm intensity, as evidenced by its lightning production capacity. Investigations suggest that the increase in thunderstorm numbers is mainly driven by increased convective available potential energy due to the increased near‐surface moisture over the TP. A decrease in ice‐phase processes and more warm rainfall across the TP, caused by increased near‐surface moisture in western TP and warming in eastern TP, has resulted in decreased thunderstorm intensity. These findings highlight the importance of considering the impact of severe weather changes on the future global energy reallocation and water cycle.

Thunderstorms are weak but frequent, and exhibit unique charge structures over the Tibetan Plateau (TP) where the average elevation is higher than 4 km. In this study, all detected thunderstorms over the TP between 1998 and 2013 by TRMM were divided into four intensity categories: weak, median, severe and extreme. This classification was based on the 75%, 90%, and 99% values of flash rate, maximum 40 dBZ height, minimum 85 GHz polarization-corrected temperature (PCT), and minimum 37 GHz PCT, respectively. The monthly distributions of thunderstorm intensity show that all categories mostly occur in summer over most regions of the TP, and in spring near the Himalayas. Although the peaks of thunderstorms occur during 1300-1600 LT, the thunderstorms occurring in the early morning and evening have a high probability of developing into severe and extreme thunderstorms. This is distinct from the thunderstorms over the Sichuan Basin, the surrounding areas, and the middle and lower reaches of the Yangtze River at the same latitude. On the basis of westerlies- and monsoon-dominated regions, as well as the altitude, the TP was divided into four regions: the eastern, northern, southern and western regions of the TP (namely ETP, NTP, STP and WTP, respectively). The ETP and STP are primarily influenced by the monsoon, with the ETP at a lower altitude than the STP. Conversely, the WTP and NTP are affected by the westerlies, with the WTP situated at a higher altitude than the NTP. Thunderstorms over the ETP are more likely to be severe and extreme than those over the NTP. The percentage of weak thunderstorms is highest over the WTP. It is found that the maximum top height, development depth, horizontal development area, and development volume at 20 dBZ, 30 dBZ, and 40 dBZ echoes are largest over the ETP, followed by the NTP and STP, while being smallest over the WTP. The results imply that thunderstorms influenced by the monsoon are larger and more likely to be severe and extreme than those influenced by the westerlies.

Natural lightning is random and transient, and precise research on the internal characteristics of lightning discharge channels requires high-resolution measurements. And the existing regenerative spectrometers have low time resolution, so it is particularly important to design a fibre-optic lightning high-speed spectrometer device with high time resolution. The article introduces a fibre-optic high-speed lightning spectrometer device, i.e. using the grating diffraction principle to decompose lightning into spectra consisting of different wavelengths, fixing the position of the bright stripe in the centre of the lightning spectrum with a spectral positioning camera, picking up the light intensity signals with N fibre-optic arrays uniformly arranged on the spectral lines of different wavelengths and transmitting the light signals to a high-speed data acquisition module, which will perform the data acquisition after photoelectric conversion. It involves the fields of optics and electricity. The device includes: a grating, a convex lens, a receiving screen, a fibre optic, a photoelectric converter, a high speed data acquisition module and a spectral positioning camera. The device can measure the high time resolution waveform variation curves of N light components of different wavelengths or frequencies of a lighting, which can be used for more detailed lighting studies.

Using the lightning VHF interferometer, three types of discharges on the preexisting negative leader channels of a positive cloud-to-ground lightning flash were observed. The first type involved small-scale cluster discharges during the simultaneous development of the upper horizontally negative leader and downward positive leader before the return stroke. These discharges exhibited similar characteristics and radiation features as the needle-like discharges on the positive leader. Over time, their occurrence positions progressed toward the head of the negative leader, and some cluster discharges had the potential to develop into new negative branches. The other two types of re-discharges occurred after the return stroke. Immediately after the return stroke, rapid discharges initiated near the head of the negative leader, developed along the preexisting negative leader channel, and caused the decayed negative leaders to progress forward again. Subsequently, numerous lateral discharges breaking down the air occurred, distributed widely throughout the negative leader channel. These discharges developed rapidly, gradually slowing down over time until the long continuous current ended. In comparison to the positive leader discharges before the return stroke, which showed no obvious recoil leader discharges, the negative leader channel was more prone to extinguish. These re-discharges on the preexisting negative leader channel were influenced by both radial and longitudinal electric fields of flash channels, and they could also generate a backward surging current wave to sustain the discharge process on the positive leader or grounded channel.