STAR Seminars

STAR Science Seminars
Presenter:
Ryo Yoshida, Satellite Program Division, Observation Department, Japan Meteorological Agency (JMA)

Sponsor(s):
STAR Science Seminar Series

Remote Access:
WebEx:
Event Number:   908 461 130   
Password: STARSeminar
Event address for attendees:
https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=m8e04fc2ec648b216893adf9e27720591

Audio:
  
+1-415-527-5035 US Toll
Access code: 908 461 130

Abstract:
The presentation will provide an overview of JMA's Himawari series satellites and results of Geostationary Lightning Mapper (GLM) product evaluation conducted in coordination with NOAA's GLM science team.
Space-based lightning mappers in geostationary orbit are expected to provide frequent and spatially extensive lightning observations over land and ocean. GLM aboard GOES-16 launched in November 2016 is the first lightning mapper in geostationary orbit, followed by GOES-17 GLM launched in March 2018.
This study validated the GLM Level 2 product using ground-based lightning observations, in terms of flash geographical distribution and flash detection efficiency, as well as group timing and geolocation accuracy. In addition, a newly developed parallax collection method using the ABI cloud top height product was applied to GLM groups, and their geolocation accuracy was compared with operational data.

Bio:

Mr. Yoshida is completing a 1-year research visit program sponsored by the Japanese Government at the NOAA Satellite and Information Service (NESDIS) Cooperative Institute for Climate and Satellites (CICS) at the University of Maryland. His research at CICS has focused on Geostationary Lightning Mapper (GLM) product evaluation and NOAA's advanced efforts on weather satellites.
Mr. Yoshida's is a Scientific Officer at the Satellite Program Division, Japan Meteorological Agency (JMA). His work at JMA has been concerned with the Himawari series satellites. He developed Himawari-8/9 image navigation and registration processing operating on the ground system. He was also responsible for development and implementation of Himawari-8/9 level 2 products at the Meteorological Satellite Center of JMA. Mr. Yoshida received B.S. and M.S. degrees in geophysics from Tohoku University, in 2007 and 2009, respectively.

POC:
Stacy Bunin, stacy.bunin@noaa.gov

STAR Science Seminars
Presenters:
Imme Ebert-Uphoff of CIRA and Elizabeth Barnes and Ben Toms of Colorado State University
(presenting remotely)

Sponsor(s):
STAR Science Seminar Series

Remote Access:
WebEx:
Event Number:    904 841 535   
Password: STARSeminar
Event address for attendees:
https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=ma4af12891805bfb155ba26ba1d4a4330

Audio:
  
    +1-415-527-5035 US Toll
    Access code: 904 841 535

Abstract:
Artificial neural networks (ANNs) have emerged as an important tool for many environmental science applications.  However, ANNs are not naturally transparent and are thus often used as a black box, i.e. without detailed understanding of their reasoning.  Fortunately, new tools for the interpretation of ANN models are becoming available from the field of explainable AI.  Such tools can provide great benefits for earth science researchers.  In this tutorial we first provide a general overview, including methods for both ANN visualization and ANN attribution.  Then we focus on one method in detail, namely layer-wise relevance propagation (LRP; sometimes known as Deep Taylor decomposition), and show how it can be used to identify the specific elements of the input that were most important for the ANN's prediction. Thus, this method helps "open the black box" and attribute specific predictions to specific predictands.  We find LRP methods to be particularly useful, yet few in the earth science community seem to have discovered them.  We demonstrate the use of LRP methods for a variety of applications related to weather and climate, and show their use for tasks ranging from debugging and designing ANN networks to gaining new scientific insights for atmospheric science applications.
About the

Presenter(s):

Imme Ebert-Uphoff received B.S. and M.S. degrees in Mathematics from the Technical University of Karlsruhe (known today as Karlsruhe Institute of Technology or KIT).  She received M.S and Ph.D. degrees in Mechanical Engineering from the Johns Hopkins University. She was a faculty member in Mechanical Engineering at Georgia Tech for over 10 years, before joining the Electrical & Computer Engineering department at Colorado State in 2011 as research professor.  Her research interests are in applying data science methods to climate applications.  She is also very involved in activities to build bridges between the AI community and the earth science community, including serving on the steering committee of the annual Climate Informatics workshop, and of the NSF sponsored research coordination network (RCN) on Intelligent Systems for the Geosciences.  Starting July 1, 2019, she is spending 50% of her time with CIRA to support their machine learning activities.
Dr. Elizabeth (Libby) Barnes is an associate professor of Atmospheric Science at Colorado State University. She joined the CSU faculty in 2013 after obtaining dual B.S. degrees (Honors) in Physics and Mathematics from the University of Minnesota, obtaining her Ph.D. in Atmospheric Science from the University of Washington, and spending a year as a NOAA Climate & Global Change Fellow at the Lamont-Doherty Earth Observatory. Professor Barnes' research is focused on large scale atmospheric variability and the data analysis tools used to understand its dynamics. Topics of interest include jet-stream dynamics, Arctic-midlatitude connections, subseasonal-to-seasonal (S2S) prediction of extreme weather events (she is currently Task Force Lead for the NOAA MAPP Subseasonal-to-Seasonal (S2S) Prediction Task Force), health-related climate impacts, and data science methods for climate research (e.g. machine learning, causal discovery). She teaches graduate courses on fundamental atmospheric dynamics and data science and statistical analysis methods.
Ben Toms is a fourth year PhD student in the Barnes research group in the Department of Atmospheric Science at Colorado State University.  His PhD research focuses on using neural networks to improve our understanding of decadal predictability within the climate system.  This research requires a fundamental understanding of neural networks and techniques for their interpretation, so he enjoys testing which methods proposed by the computer science community are transferrable to atmospheric science.

POC:
Stacy Bunin, stacy.bunin@noaa.gov

STAR Science Seminars
Presenter:
Jim Yoe, NWS/NCEP

Sponsor(s):
STAR Science Seminar Series

Remote Access:
WebEx:
Event Number:        900 946 681   
Password: STARSeminar
Event address for attendees:
    https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=m73bfc45938367d9230387b008e1bf98c

Audio:
       +1-415-527-5035 US Toll
    Access code:   900 946 681

Abstract:
TBD

Bio:

James G. (Jim) Yoe serves in the Office of the Director of the National Centers for Environmental Prediction as NCEP's Research Transition Manager.  In this capacity he coordinates NCEP's activities for the Science and Technology Integration portfolio and the Observations portfolio, and he serves as the Chief Administrative Officer of the Joint Center for Satellite Data Assimilation (JCDSA.)  Prior to joining NCEP, he spent 14 years  with the National Environmental Satellite Data and Information Service,  as a member of the NPOESS Data Exploitation Project, after working in STAR and serving as Deputy Director of the JCSDA, and developing applications for space-based remote sensors including Doppler Wind lidar and GPS Radio Occultation.  He earned BS and PhD degrees in physics from the University of the South and Clemson University, respectively, and conducted post-doctoral research investigating winds, waves, and turbulence using MST Doppler radar and UV lidar at the Max Planck Institute for Aeronomy in Germany.  His hobbies include gardening, playing the guitar, and archery. Dogs love him.

POC:
Stacy Bunin, stacy.bunin@noaa.gov

STAR Science Seminars
Presenter:
Brian Hunt, University of Maryland

Sponsor(s):
STAR Science Seminar Series

Remote Access:
WebEx:
Event Number:   900 990 334   
Password: STARSeminar
Event address for attendees:
https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=mf010724abf20590830c96cf584ac7410

Audio:
  
    +1-415-527-5035 US Toll
    Access code: 900 990 334

Abstract:
Brian will present recent work using machine learning to analyze time series data from chaotic systems.  Most of the results concern learning the systems dynamics to facilitate forecasting and climate simulation, but I will also discuss potential applications in data assimilation.  First I will show successful application of a particular form of machine learning called reservoir computing to data from relatively low-dimensional systems, and discuss a partial theory for how the method works.  Then I will present extensions of the method to handle high-dimensional, spatially-extended systems using parallel computing, and to a hybrid approach using machine learning to improve an imperfect physics-based model.

Bio:

Brian R. Hunt received a master's degree in mathematics from the University of Maryland in 1983. He went on to study applied mathematics at Stanford University, receiving a Ph.D. in 1989 for research in fluid dynamics and geometric optics. He has since returned to the University of Maryland to pursue research in dynamical systems and fractal geometry, where he is currently a Professor of Mathematics with a joint appointment in the Institute for Physical Science and Technology.

POC:
Stacy Bunin, stacy.bunin@noaa.gov