A Heterogeneous Spatiotemporal Attention Fusion Prediction Network for Precipitation Nowcasting

Niu, Dan; Che, Hongshu; Shi, Chunlei; Zang, Zengliang; Wang, Hongbin; Chen, Xunlai; Huang, Qunbo

doi:10.1109/jstars.2023.3310361

Cited by 5 publications

(1 citation statement)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The FY-3D microwave sounding instruments don't have window channels at 23.8GHz and 31.4GHz, which makes it very difficult to obtain Cloud Liquid Water (CLW) and Total Precipitation Water (TPW). But the unique 118.75 GHz channel of FY-3D happens to be very sensitive to water condensates [2][3][4]. These are the key problems to be solved in the application of FY-3D microwave data assimilation.…”

Section: Introductionmentioning

confidence: 99%

Numerical experimental research on assimilation of cloud-free radiances from fengyun-3D microwave sounding observations over land surface

Liu,

Huang,

Wang

et al. 2024

Fifth International Conference on Geology, Mapping, and Remote Sensing (ICGMRS 2024)

View full text Add to dashboard Cite

Currently, most of the domestic numerical weather prediction systems in China only assimilate satellite microwave sounding observations over sea. There is a need to further explore the potential applications of satellite observations in the sensitive sounding channels over land. This is in line with the practical operational development needs of numerical weather prediction in China, and several approaches were proposed in this paper. These approaches aimed to solve the basic problems of accurate inversion of surface emissivity and cloud detection, including a dynamic surface emissivity calculation method constrained by wavelet data set, and an intelligent cloud detection algorithm based on GPR (Gaussian Process Regression). These methods were developed to enable the practical application of assimilating land-over satellite data in the surface-sensitive sounding channels. The research results show that: The number of land-over data entering the assimilation system was found to be equivalent to the number of sea-surface observations, significantly expanding the assimilation data source. The 24-hour and 48-hour forecast effect of heavy precipitation was found to be better than that of the operational model.

show abstract

Section: Introductionmentioning

confidence: 99%

Numerical experimental research on assimilation of cloud-free radiances from fengyun-3D microwave sounding observations over land surface

Liu,

Huang,

Wang

et al. 2024

Fifth International Conference on Geology, Mapping, and Remote Sensing (ICGMRS 2024)

View full text Add to dashboard Cite

show abstract

A new strong convective precipitation forecasting method based on attention mechanism and spatio-temporal reasoning

Zhao,

Wang,

Zhao

et al. 2024

Sci Rep

View full text Add to dashboard Cite

FsrGAN: A Satellite and Radar-Based Fusion Prediction Network for Precipitation Nowcasting

Niu,

Li,

Wang

et al. 2024

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Precipitation nowcasting refers to the prediction of small-scale precipitation events at minute and kilometer scales within the upcoming 0 to 2 hours, which significantly impacts both human activities and daily life. However, prevailing deep learning models have primarily focused on a single radar echo data source, limiting their ability to effectively capture intricate and rapidly evolving precipitation patterns. Thus, meteorological satellite is considered to supplement radar echo data. To achieve a comprehensive integration of multi-source data with enhanced details, a two-stage fusion satellite and radar GAN-based prediction network (named FsrGAN) is proposed. In the first stage, we design a satellite-radar fusion prediction network known as FsrNet. This network employs an encoder-fusion-decoder architecture, where a novel spatial-channel attention (SCA) is proposed to enhance the filtering and fusion of multi-source and multi-scale features. In the second stage, we introduce a GANbased network (FusionGAN) that also mines the complementary information of satellite images to sharpen the first-stage predicted radar maps with more details. Experiments are conducted on meteorological dataset in the Yangtze River Delta (YRD) region. The test results exhibit the notably superior performance of our model in terms of image quality and precipitation forecasting metrics in comparison to traditional optical flow-based methods and some well-known deep learning methods (ConvLSTM, Con-vGRU, TrajGRU and PredRNN++). More importantly, our fusion model using satellite and radar data demonstrates the ability to predict convective initiation.

show abstract

A Heterogeneous Spatiotemporal Attention Fusion Prediction Network for Precipitation Nowcasting

Cited by 5 publications

References 48 publications

Numerical experimental research on assimilation of cloud-free radiances from fengyun-3D microwave sounding observations over land surface

Numerical experimental research on assimilation of cloud-free radiances from fengyun-3D microwave sounding observations over land surface

A new strong convective precipitation forecasting method based on attention mechanism and spatio-temporal reasoning

FsrGAN: A Satellite and Radar-Based Fusion Prediction Network for Precipitation Nowcasting

Contact Info

Product

Resources

About