Enter Note Done
Go to previous page in this tab
Session
  • Presentation | H13C: Advances in Machine Learning for Earth Science: Observation, Modeling, and Applications II Oral
  • Oral
  • Share Icon
    Map Icon
    Bookmark Icon
  • H13C-04: Fusion of Multi-Source Precipitation Records via Coordinate-Based Generative Models
    • Schedule
    Notes
  • Location IconNew Orleans Theater B
    NOLA CC
    Set Timezone
    •  
    View Map

Generic 'disconnected' Message
Author(s):
Sencan Sun, Tsinghua University (First Author, Presenting Author)
Baoxiang Pan, Institute of Atmospheric Physics, Chinese Academy of Sciences
Lu Li, Sun Yat-Sen University, School of Atmospheric Sciences
Xin Li, National Tibetan Plateau Data Center (TPDC), State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources (TPESER), Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China, China
Efi Foufoula-Georgiou, University of California Irvine
Yanluan Lin, Tsinghua University, Department of Earth System Science
Congyi Nai, Institute of Atmospheric Physics, Chinese Academy of Sciences


Measuring and predicting rainfall accurately is a major challenge for scientists. Different sources of rainfall data—such as ground-based rain gauges, satellite observations, and computer models—often give conflicting results. Each method has its own strengths and weaknesses, and no single source can be considered perfect. Yet many existing AI techniques rely on only one data type, which can lead to misleading conclusions.


Our study introduces a new framework called PRIMER (Precipitation Record Infinite MERging), which combines these different data sources into a more reliable and complete view of rainfall. Instead of ignoring uncertainty, PRIMER uses artificial intelligence to learn from disagreements between sources. It can fill in missing data, correct known errors, and even apply its knowledge to new situations without being retrained.


By blending information from satellites, ground stations, and models, PRIMER provides more consistent and accurate rainfall estimates. This has major benefits for flood forecasting, water resource management, and preparing for climate projection. Our work shows how AI can turn messy, incomplete data into useful insights for both science and society.




Scientific Discipline
Suggested Itineraries
Neighborhood
Type
Main Session
Discussion