Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM Model
Radar echo extrapolation has been widely developed in previous studies for precipitation and storm nowcasting. However, most studies have focused on two-dimensional radar images, and extrapolation of multi-altitude radar images, which can provide more informative and visual forecasts about weather s...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-08-01
|
Series: | Remote Sensing |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-4292/14/17/4256 |
_version_ | 1797493340960194560 |
---|---|
author | Nengli Sun Zeming Zhou Qian Li Jinrui Jing |
author_facet | Nengli Sun Zeming Zhou Qian Li Jinrui Jing |
author_sort | Nengli Sun |
collection | DOAJ |
description | Radar echo extrapolation has been widely developed in previous studies for precipitation and storm nowcasting. However, most studies have focused on two-dimensional radar images, and extrapolation of multi-altitude radar images, which can provide more informative and visual forecasts about weather systems in realistic space, has been less explored. Thus, this paper proposes a 3D-convolutional long short-term memory (ConvLSTM)-based model to perform three-dimensional gridded radar echo extrapolation for severe storm nowcasting. First, a 3D-convolutional neural network (CNN) is used to extract the 3D spatial features of each input grid radar volume. Then, 3D-ConvLSTM layers are leveraged to model the spatial–temporal relationship between the extracted 3D features and recursively generate the 3D hidden states correlated to the future. Nowcasting results are obtained after applying another 3D-CNN to up-sample the generated 3D hidden states. Comparative experiments were conducted on a public National Center for Atmospheric Research Data Archive dataset with a 3D optical flow method and other deep-learning-based models. Quantitative evaluations demonstrate that the proposed 3D-ConvLSTM-based model achieves better overall and longer-term performance for storms with reflectivity values above 35 and 45 dBZ. In addition, case studies qualitatively demonstrate that the proposed model predicts more realistic storm evolution and can facilitate early warning regarding impending severe storms. |
first_indexed | 2024-03-10T01:18:38Z |
format | Article |
id | doaj.art-02461a8a3887466ebc70ae76e50b26aa |
institution | Directory Open Access Journal |
issn | 2072-4292 |
language | English |
last_indexed | 2024-03-10T01:18:38Z |
publishDate | 2022-08-01 |
publisher | MDPI AG |
record_format | Article |
series | Remote Sensing |
spelling | doaj.art-02461a8a3887466ebc70ae76e50b26aa2023-11-23T14:03:33ZengMDPI AGRemote Sensing2072-42922022-08-011417425610.3390/rs14174256Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM ModelNengli Sun0Zeming Zhou1Qian Li2Jinrui Jing3The College of Meteorology and Oceanography, National University of Defense Technology, Changsha 410005, ChinaThe College of Meteorology and Oceanography, National University of Defense Technology, Changsha 410005, ChinaThe College of Meteorology and Oceanography, National University of Defense Technology, Changsha 410005, ChinaThe College of Meteorology and Oceanography, National University of Defense Technology, Changsha 410005, ChinaRadar echo extrapolation has been widely developed in previous studies for precipitation and storm nowcasting. However, most studies have focused on two-dimensional radar images, and extrapolation of multi-altitude radar images, which can provide more informative and visual forecasts about weather systems in realistic space, has been less explored. Thus, this paper proposes a 3D-convolutional long short-term memory (ConvLSTM)-based model to perform three-dimensional gridded radar echo extrapolation for severe storm nowcasting. First, a 3D-convolutional neural network (CNN) is used to extract the 3D spatial features of each input grid radar volume. Then, 3D-ConvLSTM layers are leveraged to model the spatial–temporal relationship between the extracted 3D features and recursively generate the 3D hidden states correlated to the future. Nowcasting results are obtained after applying another 3D-CNN to up-sample the generated 3D hidden states. Comparative experiments were conducted on a public National Center for Atmospheric Research Data Archive dataset with a 3D optical flow method and other deep-learning-based models. Quantitative evaluations demonstrate that the proposed 3D-ConvLSTM-based model achieves better overall and longer-term performance for storms with reflectivity values above 35 and 45 dBZ. In addition, case studies qualitatively demonstrate that the proposed model predicts more realistic storm evolution and can facilitate early warning regarding impending severe storms.https://www.mdpi.com/2072-4292/14/17/4256convective storm nowcasting3D radar echo extrapolationdeep learning3D spatial features3D-ConvLSTM |
spellingShingle | Nengli Sun Zeming Zhou Qian Li Jinrui Jing Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM Model Remote Sensing convective storm nowcasting 3D radar echo extrapolation deep learning 3D spatial features 3D-ConvLSTM |
title | Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM Model |
title_full | Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM Model |
title_fullStr | Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM Model |
title_full_unstemmed | Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM Model |
title_short | Three-Dimensional Gridded Radar Echo Extrapolation for Convective Storm Nowcasting Based on 3D-ConvLSTM Model |
title_sort | three dimensional gridded radar echo extrapolation for convective storm nowcasting based on 3d convlstm model |
topic | convective storm nowcasting 3D radar echo extrapolation deep learning 3D spatial features 3D-ConvLSTM |
url | https://www.mdpi.com/2072-4292/14/17/4256 |
work_keys_str_mv | AT nenglisun threedimensionalgriddedradarechoextrapolationforconvectivestormnowcastingbasedon3dconvlstmmodel AT zemingzhou threedimensionalgriddedradarechoextrapolationforconvectivestormnowcastingbasedon3dconvlstmmodel AT qianli threedimensionalgriddedradarechoextrapolationforconvectivestormnowcastingbasedon3dconvlstmmodel AT jinruijing threedimensionalgriddedradarechoextrapolationforconvectivestormnowcastingbasedon3dconvlstmmodel |