Presented By: Climate and Space Sciences and Engineering
CLASP Seminar Series: Prof. Zhaoxia Pu of the University of Utah
Prof. Zhaoxia Pu of the University of Utah will give a lecture as part of the CLASP Seminar Series. Please join us!
Prof. Pu’s presentation is titled "Data Assimilation and Machine Learning at the Frontier of Earth System Modeling and Prediction" and will take place on Thursday, November 4 at 3:30 p.m. EDT.
This seminar will be in person at the Climate and Space Research Building Auditorium, Room 2246.
The seminar may also be viewed via Zoom.
Please contact lhopkins@umich to request zoom access.
ABSTRACT:
Data assimilation combines current available observations with numerical simulations constrained by previous observations to obtain the best possible analysis and prediction of the Earth system. It plays a central role in modern numerical weather and climate prediction. Due to its ability to integrate all sources of observations and Earth system modeling information, data assimilation has also been a powerful tool for model parameter estimation and optimization, reanalysis, model error estimation and correction, coupled Earth system modeling, ensemble forecasting, and observing system development.
With remarkable advances in machine learning and deep learning (ML/DL) techniques, there is rapidly growing interest in applying ML/DL in Earth system modeling and prediction. While ML/DL provides different ways to deal with the information and modeling, it shares similar aims with data assimilation: to learn about the world using observations. This similarity makes it conceptually feasible to integrate standard ML/DL techniques and ideas into data assimilation workflows for improved Earth system modeling and prediction.
This seminar emphasizes recent accomplishments in atmospheric data assimilation with the most recent research results on satellite and radar data assimilation for improved severe weather forecasting, coupled land-atmosphere data assimilation for enhanced representation of land-atmosphere interactions, and data assimilation in support of observing system development. The recent application of ML/DL in weather and climate prediction will be reviewed. The future direction of Earth system modeling and prediction with combined data assimilation and ML/DL methods will be discussed, particularly in the context of predicting weather and climate extremes, as noted in the recent United Nations Intergovernmental Panel on Climate Change report.
Prof. Pu’s presentation is titled "Data Assimilation and Machine Learning at the Frontier of Earth System Modeling and Prediction" and will take place on Thursday, November 4 at 3:30 p.m. EDT.
This seminar will be in person at the Climate and Space Research Building Auditorium, Room 2246.
The seminar may also be viewed via Zoom.
Please contact lhopkins@umich to request zoom access.
ABSTRACT:
Data assimilation combines current available observations with numerical simulations constrained by previous observations to obtain the best possible analysis and prediction of the Earth system. It plays a central role in modern numerical weather and climate prediction. Due to its ability to integrate all sources of observations and Earth system modeling information, data assimilation has also been a powerful tool for model parameter estimation and optimization, reanalysis, model error estimation and correction, coupled Earth system modeling, ensemble forecasting, and observing system development.
With remarkable advances in machine learning and deep learning (ML/DL) techniques, there is rapidly growing interest in applying ML/DL in Earth system modeling and prediction. While ML/DL provides different ways to deal with the information and modeling, it shares similar aims with data assimilation: to learn about the world using observations. This similarity makes it conceptually feasible to integrate standard ML/DL techniques and ideas into data assimilation workflows for improved Earth system modeling and prediction.
This seminar emphasizes recent accomplishments in atmospheric data assimilation with the most recent research results on satellite and radar data assimilation for improved severe weather forecasting, coupled land-atmosphere data assimilation for enhanced representation of land-atmosphere interactions, and data assimilation in support of observing system development. The recent application of ML/DL in weather and climate prediction will be reviewed. The future direction of Earth system modeling and prediction with combined data assimilation and ML/DL methods will be discussed, particularly in the context of predicting weather and climate extremes, as noted in the recent United Nations Intergovernmental Panel on Climate Change report.
Explore Similar Events
-
Loading Similar Events...
