AI Forecast of Extreme Rainfall in South China
Project Description
Severe rainstorms in South China present a significant forecasting challenge, often causing widespread disruption and economic impact. While traditional numerical weather prediction models are foundational, they can struggle to accurately capture the rapid onset and localized intensity of these extreme events. This project tackles this critical gap by harnessing the power of artificial intelligence. By leveraging extensive meteorological observation datasets, we are developing advanced deep learning models capable of predicting extreme rainfall with a crucial lead time of one to three days. This research not only pushes the boundaries of AI applications in atmospheric science but also aims to directly improve early warning systems and enhance regional climate resilience.
Supervisor
SHI, Xiaoming
Quota
3
Course type
UROP1100
UROP2100
UROP3100
UROP3200
UROP4100
Applicant's Roles
1) Data Curation & Processing: Extract, clean, and quality-control large volumes of regional meteorological observation data to construct robust training datasets.
2) AI Model Training: Assist in developing, training, and fine-tuning deep learning architectures (utilizing modern frameworks such as PyTorch, JAX, or Flax) tailored for spatial-temporal weather forecasting.
3) Performance Evaluation: Test and validate the AI models against historical rainfall events, analyzing the outputs to identify strengths and areas for algorithmic improvement.
4) Scientific Communication: Visualize complex forecasting results, document methodologies, and actively contribute to research group discussions.
2) AI Model Training: Assist in developing, training, and fine-tuning deep learning architectures (utilizing modern frameworks such as PyTorch, JAX, or Flax) tailored for spatial-temporal weather forecasting.
3) Performance Evaluation: Test and validate the AI models against historical rainfall events, analyzing the outputs to identify strengths and areas for algorithmic improvement.
4) Scientific Communication: Visualize complex forecasting results, document methodologies, and actively contribute to research group discussions.
Applicant's Learning Objectives
1) Applied AI & Technical Proficiency: Gain practical, hands-on experience building and deploying machine learning models using high-performance computing tools in a real-world scientific context.
2) Interdisciplinary Domain Knowledge: Develop a solid foundational understanding of atmospheric science, specifically focusing on regional meteorology and the dynamics of extreme weather in South China.
3) Analytical Problem Solving: Learn the complete data-to-discovery pipeline, translating raw, complex environmental data into structured, predictive algorithms.
4) Research Readiness: Cultivate critical thinking and scientific communication skills, building a strong foundation for future graduate studies or careers in environmental tech and research.
2) Interdisciplinary Domain Knowledge: Develop a solid foundational understanding of atmospheric science, specifically focusing on regional meteorology and the dynamics of extreme weather in South China.
3) Analytical Problem Solving: Learn the complete data-to-discovery pipeline, translating raw, complex environmental data into structured, predictive algorithms.
4) Research Readiness: Cultivate critical thinking and scientific communication skills, building a strong foundation for future graduate studies or careers in environmental tech and research.
Complexity of the project
Challenging