Enabling Petascale Ensemble-based Data Assimilation for Numerical Analysis and Prediction of High-Impact Weather

Project Details

Project Lead
Andy Li 
Project Manager
Andy Li 
Project Members
Min Li, Xin Yang, Han Zhao, Rui Yang, Ze Yu, Yuhui Wang, Cairong Yan, Shivam Tiwari  
Institution
University of Florida, Department of Electrical and Computer Engineering  
Discipline
Computer Science (401) 
Subdiscipline
40.04 Atmospheric Sciences and Meteorology 

Abstract

This proposal addresses the most challenging problems of very-high-resolution Numerical Weather Prediction, obtaining the optimal state estimations for initializing ensembles of predictions by assimilating the highest volume of weather observations available, and addressing problem sizes and scales that are only attainable on petascale computing platforms. The project is focused on the very hard problem of thunderstorm initiation, propagation and evolution. The effect of this investment in more accurate and timely forecast of thunderstorms can have a potentially transformative impact on hazard analysis and management.

Intellectual Merit

This project investigates ensemble Kalman filter based data assimilation methods to predict severe weather phenomena at finer resolutions at scale. A software toolkit that eases the development of sophisticated numerical methods for such weather forecasting applications will be developed.

Broader Impacts

Severe weather (such storms and tornados) affects our life and society. This project is positioned to leverage petascale computing facilities to tackle such long-lasting problems at accurate weather forecasting for narrow-region localized weather events (e.g. storms and tornados) at finer grains (< 1 km). Through the interdisciplinary projects, graduate and undergraduate students will be well trained. Education outreach is under plan too.

Scale of Use

We will start with small scales (tens to 128 cores), and will request for hundreds of cores when we pass the preliminary stage.