General capability of an adaptive learning model for dual-polarization radar rainfall mapping

Radar quantitative precipitation estimation (QPE) plays an important role in water, weather, and climate research due to the advantages of radars in providing seamless scan of precipitation in both time and spatial domains. With the development of dual-polarization radar systems and the innovation o...

Full description

Saved in:
Bibliographic Details
Published inIEEE International Geoscience and Remote Sensing Symposium proceedings pp. 5567 - 5570
Main Authors Wang, Liangwei, Chen, Haonan
Format Conference Proceeding
LanguageEnglish
Published IEEE 07.07.2024
Subjects
Adaptation models
Adaptive learning
Deep learning
Meteorological radar
quantitative precipitation estimation
Radar remote sensing
Rain
Technological innovation
weather radar
Online AccessGet full text

Cover

More Information
Summary:Radar quantitative precipitation estimation (QPE) plays an important role in water, weather, and climate research due to the advantages of radars in providing seamless scan of precipitation in both time and spatial domains. With the development of dual-polarization radar systems and the innovation of precipitation estimation algorithms based on deep learning, the accuracy of radar QPE has been significantly improved. However, the deep learning QPE models trained using local data may still be subject to degraded performance when applied in other regions. This paper quantifies the generalization capability of a deep learning QPE model trained using data in Florida, U.S. over other regions such as Oklahoma, U.S., which is characterized by different precipitation characteristics.
ISSN:2153-7003
DOI:10.1109/IGARSS53475.2024.10642359