RDA-Unet: A Retrieval Model for Radar Composite Reflectivity Factor from Himawari-8 Observations
Lists
Only you see this
Abstract
Due to the unbalanced and insufficient distribution of meteorological radar in China, there is a large blind area of radar network in the southwest and the maritime region. With its larger observation range and less error-prone observation results, geostationary meteorological satellite can compensate for the radar network gap to a certain extent, however, the existing retrieval models are difficult to extract the complex multi-channel satellite data features effectively for the channel and spatial dimensions. To address the above problems, a U-Net-based network architecture called RDA-Unet is proposed. Specifically, the spatial attention module and channel attention module are introduced to aggregate features related to radar echoes in both spatial and channel dimensions in the original U-net architecture. Finally, the comparison experiments of multiple indicators verify that the algorithm in this paper has certain improvement in the accuracy of echo retrieval of different sizes, which demonstrates the superiority of the algorithm.
BibTeX
@inproceedings{jin2023rda,
author = {Zhengyong Jin and Huichao Lin and Xiaolong Xu},
title = {{RDA-Unet: A Retrieval Model for Radar Composite Reflectivity Factor from Himawari-8 Observations}},
booktitle = {HPCC/DSS/SmartCity/DependSys},
year = {2023},
doi = {10.1109/hpcc-dss-smartcity-dependsys60770.2023.00143},
}Loading related content...