A satellite system brings stronger rainfall detection into real time
In plain words
Satellite pictures show clouds, but estimating how much rain reaches the ground is difficult. Fast rainfall products may be less accurate, while stronger products can arrive months later.
This paper introduces Huayu-Global, which uses three satellite streams to make a shared rainfall map every half hour. In comparisons with ground stations outside its training years, it detected rain better overall than the seven other tested products. The result applies to the stations and periods measured.
The research
Can satellite rainfall estimates be available quickly enough for time-sensitive decisions while still detecting precipitation well? Prompt products can have weaker detection skill, while higher-quality products may arrive months later. Information delivered after a decision window cannot support actions such as flood warning.
Huayu-Global combines three satellite-specific rainfall models into a half-hourly map, then compares its estimates with independent station records.
Researchers asked whether a prompt satellite product could detect precipitation as well as products that arrive later. They built Huayu-Global to combine three satellite streams in a shared, half-hourly precipitation field, then compared it with seven products using independent station observations.
Across matched three-hour records from two periods outside the training years, Huayu-Global had the highest critical success index, a measure of correctly detected rain events that counts misses and false alarms.
Its score exceeded even delayed IMERG Final Run, supporting the authors’ conclusion that its station-based skill can exceed that product on reported measures.
The evaluation uses an unevenly distributed land station network and does not directly test open-ocean precipitation. The paper’s “real-time” label describes when inputs are available and local processing time; it does not measure the time until public release.
Analysis
Deep
- Original paper authors
- Zijiang Song, Shuang Yang, Yuying Li, Zilong Zhu, Xigang Sun
- Submitted to arXiv
- License
- arXiv non-exclusive