[Eoas-seminar] MET seminar at 3:30 PM on Thursday, Mar. 25, 2021

[eoas-seminar at lists.fsu.edu]

Just a correction, that the MET seminar given by Dr. Zane Martin is this Thursday, March 25 at 3:30 PM. I know that there was a lot of interest in machine learning following one of our colloquia last year, so I hope to see a good turnout this week for Zane’s talk!

Graduate students, please stick around after the seminar for a student-only Q&A with the speaker. Zane is a postdoc at Colorado State who I’ve known since he started graduate school and he is super friendly and fun to talk to. In addition to his work in modeling of tropical dynamics, Zane has participated in a couple of field campaigns.

Cheers,

Allison

——————————————————
Allison Wing, Ph.D.
Assistant Professor
Earth, Ocean and Atmospheric Science
Florida State University
awing at fsu.edu<mailto:awing at fsu.edu>



On Mar 22, 2021, at 9:32 AM, eoas-seminar--- via Eoas-seminar <eoas-seminar at lists.fsu.edu<mailto:eoas-seminar at lists.fsu.edu>> wrote:

Hi all,

Here is an announcement that we have a MET seminar at 3:30 PM on Thursday, Mar. 25, 2021. The related information can be found in the following and the attached flyer.

Speaker: Dr. Zane K. Martin, Department of Atmospheric Sciences, Colorado State University

Title: Predicting the Madden-Julian oscillation using interpretable machine learning

Abstract:  The Madden-Julian oscillation (MJO) is among the most important modes of tropical variability on the planet, and a dominant driver of subseasonal-to-seasonal prediction skill and predictability globally. The past decade has seen substantial advances in MJO prediction using dynamical forecast models, which now show higher skill than statistical MJO forecasts. Also in recent years, an increasing body of literature has demonstrated that machine learning methods represent a new frontier in Earth science with a wide range of applications. After a brief overview of the current state of MJO prediction, we discuss how state-of-the-art machine learning can be used to make real-time MJO forecasts. We introduce a particular type of machine learning model called a neural network, and then demonstrate how it can be used to predict MJO. We show that machine learning models have high skill relative to statistical models overall, but still underperform the very best dynamical MJO models. We close by discussing the strengths of these models and how they might be used and improved going forward, including their potential to lead to insights about the MJO. We also discuss cutting-edge techniques from the field of interpretable AI that allow us to visualize how these neural network makes predictions.

Time: 3:30 PM, Thursday, Mar. 25, 2021

Zoom Meeting: https://fsu.zoom.us/j/97840279436?pwd=YzlkdnNqZG1GaDhVMnJzSmZIb2VwQT09

We will start the zoom meeting site to meet the speaker at 3:00 PM. It is also noted a post-seminar student-speaker session will start immediately after the seminar.

Cheers,

Zhaohua

<MET_Seminar_Flyer_Martin.pdf>_______________________________________________
Eoas-seminar mailing list
Eoas-seminar at lists.fsu.edu<mailto:Eoas-seminar at lists.fsu.edu>
https://lists.fsu.edu/mailman/listinfo/eoas-seminar

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.fsu.edu/pipermail/eoas-seminar/attachments/20210322/74ea3f08/attachment.html>