Deep Learning-Based Radar Composite Reflectivity Factor Estimations from Fengyun-4A Geostationary Satellite Observations

Fenglin Sun,Bo Li,Min Min,Danyu Qin
DOI: https://doi.org/10.3390/rs13112229
IF: 5
2021-06-07
Remote Sensing
Abstract:Ground-based weather radar data plays an essential role in monitoring severe convective weather. The detection of such weather systems in time is critical for saving people’s lives and property. However, the limited spatial coverage of radars over the ocean and mountainous regions greatly limits their effective application. In this study, we propose a novel framework of a deep learning-based model to retrieve the radar composite reflectivity factor (RCRF) maps from the Fengyun-4A new-generation geostationary satellite data. The suggested framework consists of three main processes, i.e., satellite and radar data preprocessing, the deep learning-based regression model for retrieving the RCRF maps, as well as the testing and validation of the model. In addition, three typical cases are also analyzed and studied, including a cluster of rapidly developing convective cells, a Northeast China cold vortex, and the Super Typhoon Haishen. Compared with the high-quality precipitation rate products from the integrated Multi-satellite Retrievals for Global Precipitation Measurement, it is found that the retrieved RCRF maps are in good agreement with the precipitation pattern. The statistical results show that retrieved RCRF maps have an R-square of 0.88-0.96, a mean absolute error of 0.3-0.6 dBZ, and a root-mean-square error of 1.2-2.4 dBZ.
environmental sciences,imaging science & photographic technology,remote sensing,geosciences, multidisciplinary
What problem does this paper attempt to address?
The paper mainly addresses the following issues: ### Research Background and Objectives - **Problem Statement**: Ground-based weather radar data play a crucial role in monitoring severe convective weather (such as strong winds, heavy rain, and flash floods), but radar coverage is limited, especially in oceans and mountainous areas, which greatly restricts its effective application. ### Solution - **Research Method**: A new framework based on deep learning is proposed to retrieve Radar Composite Reflectivity Factor (RCRF) maps from observations of the new generation geostationary satellite Fengyun-4A. ### Main Steps 1. **Data Preprocessing**: Includes preprocessing of satellite and radar data. 2. **Deep Learning Regression Model**: Used to retrieve RCRF maps. 3. **Testing and Validation**: The model is tested and validated. ### Case Analysis - **Rapidly Developing Convective Cells**: Analyzed cases of rapidly developing convective cell clusters. - **Northeast Cold Vortex**: Studied a case of a northeast cold vortex. - **Super Typhoon Haishen**: Analyzed the situation of Super Typhoon Haishen. ### Results Validation - **Statistical Results**: Validation results show that the retrieved RCRF maps are in good agreement with precipitation patterns, with high correlation coefficients (0.88-0.96), mean absolute errors (0.3-0.6 dBZ), and root mean square errors (1.2-2.4 dBZ). ### Technical Details - **Data Sources**: Used data from the Fengyun-4A satellite, ground-based weather radar data, and data from the Global Precipitation Measurement (GPM) mission. - **Model Architecture**: Improved based on the U-net architecture to meet the needs of reflectivity factor estimation. - **Network Training**: Used Adaptive Moment Estimation in PyTorch for network training and defined a loss function to optimize model parameters. ### Conclusion - This study developed a deep learning-based method that can effectively retrieve Radar Composite Reflectivity Factor maps from Fengyun-4A satellite data, providing high-quality prediction results even in areas not covered by radar. This method is of great significance for improving the monitoring and early warning capabilities of severe convective weather.