A deep learning framework for lightning forecasting with multi‐source spatiotemporal data

Geng, Yangli-ao; Li, Qingyong; Lin, Tianyang; Yao, Wen; Xu, Liangtao; Zheng, Dong; Zhou, Xinyuan; Zheng, Liming; Lyu, Weitao; Zhang, Yijun

doi:10.1002/qj.4167

Cited by 28 publications

(12 citation statements)

References 39 publications

(53 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2017, the same authors expanded on their work by incorporating gated recurrent units (GRUs) and introducing the convolutional gated recurrent unit (ConvGRU) and trajectory GRU (Traj-GRU) algorithms [47]. Building upon Shi et al's foundation, Geng et al proposed LightNet, a model that utilizes numerical models (WRF) and real-time lightning observations to predict lightning occurrences in the next 0-6 hours [48,49]. In 2022, a novel approach, Seamless Lightning Nowcasting with RNN, emerged.…”

Section: Introductionmentioning

confidence: 99%

A spatio-temporal fusion deep learning network with application to lightning nowcasting

Zhou,

Fan,

Neri

2024

ICA

View full text Add to dashboard Cite

Lightning is a rapidly evolving phenomenon, exhibiting both mesoscale and microscale characteristics. Its prediction significantly relies on timely and accurate data observation. With the implementation of new generation weather radar systems and lightning detection networks, radar reflectivity image products, and lightning observation data are becoming increasingly abundant. Research focus has shifted towards lightning nowcasting (prediction of imminent events), utilizing deep learning (DL) methods to extract lightning features from very large data sets. In this paper, we propose a novel spatio-temporal fusion deep learning lightning nowcasting network (STF-LightNet) for lightning nowcasting. The network is based on a 3-dimensional U-Net architecture with encoder-decoder blocks and adopts a structure of multiple branches as well as the main path for the encoder block. To address the challenges of feature extraction and fusion of multi-source data, multiple branches are used to extract different data features independently, and the main path fuses these features. Additionally, a spatial attention (SA) module is added to each branch and the main path to automatically identify lightning areas and enhance their features. The main path fusion is conducted in two steps: the first step fuses features from the branches, and the second fuses features from the previous and current levels of the main path using two different methodsthe weighted summation fusion method and the attention gate fusion method. To overcome the sparsity of lightning observations, we employ an inverse frequency weighted cross-entropy loss function. Finally, STF-LightNet is trained using observations from the previous half hour to predict lightning in the next hour. The outcomes illustrate that the fusion of both the multi-branch and main path structures enhances the network’s ability to effectively integrate features from diverse data sources. Attention mechanisms and fusion modules allow the network to capture more detailed features in the images.

show abstract

Section: Introductionmentioning

confidence: 99%

A spatio-temporal fusion deep learning network with application to lightning nowcasting

Zhou,

Fan,

Neri

2024

ICA

View full text Add to dashboard Cite

show abstract

“…In the recent years, deep learning (DL) has been increasingly adopted in thunderstorm nowcasting algorithms (Cuomo & Chandrasekar, 2022; Geng et al., 2021; Pan et al., 2021; Zhou et al., 2020). Leinonen et al.…”

Section: Introductionmentioning

confidence: 99%

Thunderstorm Nowcasting With Deep Learning: A Multi‐Hazard Data Fusion Model

Leinonen

Hamann

Sideris

et al. 2023

Geophysical Research Letters

View full text Add to dashboard Cite

Thunderstorms frequently endanger lives and cause damage to property through several distinct physical processes. Lightning are directly hazardous to humans, igniting fires and disrupting electric infrastructure and aviation (Holle, 2014(Holle, , 2016. The heavy rain frequently produced by these storms can cause flash floods and trigger landslides (Davis, 2001;Smith et al., 1996). Thunderstorms also often produce hail, which can damage buildings, vehicles, crops and forests and is particularly dangerous to flying aircraft (Battaglia et al., 2019;Hohl et al., 2002). They are furthermore frequently associated with strong, gusty wind (Allen & Allen, 2016;Dotzek, 2003;Kelly et al., 1985). Thunderstorms are relatively compact objects that develop rapidly, and consequently, the effects of these hazards are sudden and highly localized. Timely, accurate and actionable short-term predictions of them are needed to issue early warnings to the general population, emergency services and infrastructure operators.Numerical weather prediction (NWP) can forecast thunderstorm occurrence and their typical intensity level in a wider area, but have difficulty predicting the exact time and location where thunderstorms will appear (Sun et al., 2014). On the very short term of minutes to a few hours, one can instead use nowcasting: statistical prediction using the latest observational data. In contrast to NWP models that take hours to run, nowcasting models are typically designed to produce a prediction in seconds, making them suitable for time-critical short-term warning systems. The advantages of the two approaches can also be combined in a seamless forecast. As nowcasting systems tend to be accurate on the short term but lose predictive power faster than NWP models, seamless forecasts typically use nowcasting on the short term, NWP forecasts on the long term, and a combination of the two in between (Nerini, 2019;Nerini et al., 2019).In the recent years, deep learning (DL) has been increasingly adopted in thunderstorm nowcasting algorithms

show abstract

“…The remarkable achievements of deep neural network technology in intelligent answering, weather forecasting, and voiceprint comparison reflect its advantages in solving feature fitting, pattern recognition, and multi-dimensional classification problems [14][15][16][17][18]. Therefore, many scholars apply it to wireless communication, which opens up a new way to improve the ability of signal modulation recognition.…”

Section: Introductionmentioning

confidence: 99%

An improved CNN‐LSTM network for modulation identification relying on periodic features of signal

Zhou,

Li,

Wang

2023

IET Communications

View full text Add to dashboard Cite

Aiming at the problem that modulated signals' classification and recognition accuracy is not high in the complex and changeable electromagnetic environment, this paper improves the CNN‐LSTM network and proposes a modulation recognition method of the CNN‐LSTM network (Intra‐InterNet) based on the periodic characteristics of the signal. The method uses the feature extraction ability of the convolutional neural network to extract the correlation features within the cycle and between adjacent cycles of different kinds of modulated signals. At the same time, the relevant features of the signals of non‐adjacent cycles are extracted by using the characteristics of the LSTM. This feature extraction method effectively enhances the feature extraction capability of the network by utilizing the periodicity of the modulated signal. Through the identification and verification of various modulation signals, the experimental results show that: under the signal‐to‐noise ratio of −6 dB ∼ 6 dB, the average identification accuracy rate reaches 87.29%, which effectively improves the accuracy of signal modulation mode identification.

show abstract

A deep learning framework for lightning forecasting with multi‐source spatiotemporal data

Cited by 28 publications

References 39 publications

A spatio-temporal fusion deep learning network with application to lightning nowcasting

A spatio-temporal fusion deep learning network with application to lightning nowcasting

Thunderstorm Nowcasting With Deep Learning: A Multi‐Hazard Data Fusion Model

An improved CNN‐LSTM network for modulation identification relying on periodic features of signal

Contact Info

Product

Resources

About