STAR Seminars

STAR Science Seminars

Presenter:
Eric Wendoloski, Connor Sprague, and Ingrid Guch of The Aerospace Corporation

Sponsor:
STAR Science Seminar Series

Remote Access:
WebEx
Event Number: 906 823 554  
Password: STARSeminar

Event address for attendees: https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=m85e78702a2eeab99a81b293480679bab

Audio:  
USA participants: 866-832-9297
Passcode:  6070416

Abstract:
The number of machine-learning (ML) applications has surged within the meteorological community over the last several years. This surge includes the development and application of numerous ML techniques to improve forecasting as well as physical models while reducing computational complexity and time. Given the vast trove of available satellite-based weather imagery and the gridded structure of many meteorological datasets, deep-learning (DL) methods for providing predictions and diagnostics for numerous subdomains are experiencing increased adoption. However, full adoption will require forecasters and decision makers to interpret why model output is produced given the input, especially if the output has implications for human well-being. Due to their complex architectures, interpreting DL models can be especially difficult, and models are often treated as black boxes. This work examines contemporary methods for assessing the interpretability of a convolutional neural network (CNN) trained to predict tropical cyclone (TC) intensity based on available satellite-weather data, primarily in the IR band. CNNs excel at distilling image data into the most important feature abstractions for developing functional associations between input images and required prediction output. The goal of this work is not necessarily to produce the most accurate TC intensity model, but to assess whether such a DL architecture is capable of learning physically relevant abstractions for the problem at hand. We will describe and apply interpretability methods to the TC intensity CNN model to assess the importance of physical concepts to final predictions. We will also assess the traceability of predictions across the learned network.

About the Speaker:

Eric Wendoloski is a Senior Member of the Technical staff for The Aerospace Corporation in Chantilly, VA. Since joining Aerospace, Eric has supported projects ranging from the deployment of machine-learning-based applications to large scale systems engineering studies including the NOAA Satellite Observing System Architecture (NSOSA) study. Prior to joining Aerospace, Eric obtained his B.S. (2013) in meteorology from Millersville University and his M.S. (2015) in meteorology from Penn State University. Eric was also a recipient of the Ernest F. Hollings scholarship as an undergraduate. Connor Sprague is a graduate intern at The Aerospace Corporation. His research focuses on applicable machine learning and data science for hurricane prediction and tracking, data fusion, and predictive weather algorithms. He is finishing M.S in systems engineering at the Missouri University of Science and Technology. Ingrid Guch is a Systems Director at The Aerospace Corporation. She has over 20 years' experience working with NOAA/NESDIS including in product operations, systems development, applications and research. She is currently co-located with the NOAA/NESDIS Center for Satellite Applications and Research in College Park, MD. She obtained her B.S. in mathematical sciences from University of California Santa Barbara and her M.S. in atmospheric sciences from Colorado State University.

STAR Science Seminars
with SOCD / NOAA Ocean Color Coordinating Group


This seminar was originally scheduled for 12/5/2018.

Presenter:
Dr. Timothy S. Moore - Ocean Process Analysis Laboratory
Institute for Earth, Oceans and Space
University of New Hampshire

Sponsor:

SOCD / NOAA Ocean Color Coordinating Group
The NOCCG is a NOAA organization founded in 2011 by Dr. Paul DiGiacomo, Chief of the Satellite Oceanography and Climatology Division at NOAA/NESDIS/STAR.  The purpose of the NOCCG is to keep members up to date about developments in the field of satellite ocean color and connect ocean color science development with users and applications.  We have representatives from all the NOAA line offices, including National Marine Fisheries Service, Office of Oceanic and Atmospheric Research, National Ocean Service, National Weather Service and from several levels of the National Environmental and Satellite Data and Information Service (where Paul is housed).  Dr. Cara Wilson of South East Fisheries Science Center is our current chair. We meet bi-weekly on Wednesday afternoons, 3 PM Eastern Time in room 3555 at the National Center for Weather and Climate Prediction building in College Park, MD with teleconferencing and Webex for out of town members and guests.  We host a guest speaker, usually about once a month.

Remote Access:
WebEx:
Event Number:    907 721 095
Password: NOCCG
Event address for attendees:

https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=mf7555abe45f0f5eb592ac618bd9b38b1


Audio:  
USA participants: 866-564-7828 Passcode: 9942991



Abstract:


 In the summer of 2016, a robotic sun photometer called the Sea-Viewing Wide Field-of-View Sensor (SeaWiFS) Photometer Revision for Incident Surface Measurements (SeaPRISM), was deployed at a Coast Guard channel marker in western Lake Erie, measuring atmospheric properties and spectral water-leaving radiance. The instrument was deployed by the National Oceanic and Atmospheric Administration (NOAA) to support remote sensing validation and harmful algal bloom (HAB) satellite products. The Lake Erie SeaPRISM is also part of the international federated AERONET program maintained by the National Aeronautics and Space Administration (NASA), and more specifically is part of the AERONET Ocean Color (AERNOET-OC) network. The main purpose of this component of AERONET is specific to calibration/validation efforts for ocean color. In the summer of 2017, a new 12-channel version was deployed at the same site with additional channels in the red and near-infrared. This unit is the first ‘lake' version of the SeaPRISM. In this webinar, the data collected by the SeaPRISM at this site over the last three years (2016-2018) will be examined in the context of HABs and remote sensing validation. The SeaPRISM observations in relation to remote sensing validation and on cyanobacteria blooms from hourly to weekly time scales will be highlighted in this presentation.

About the Speaker:

Dr. Moore has been working with ocean color remote sensing for over 25 years. Throughout that time, he has been involved with bio-optical algorithm development, application, and satellite validation. He was worked with ocean color imagery in marine and freshwater systems. He was a member of the NASA MODIS Science Team and NASA PACE Science Team. For the past six years, he has been working extensively in the western Lake Erie system collaborating with NOAA GLERL and other regional entities. Under a collaborative project between NOAA NESDIS, NOAA GLERL and UNH, Dr. Moore led the effort to introduce an autonomous, robotic radiometer to Lake Erie with a unique band configuration, which will be the subject of his presentation. 


POC:
Nolvia Herrera, 301-683-3308, Nolvia.Herrera@noaa.gov
NOCCG Coordinator: Veronica P. Lance, PhD, NOAA, 301-683-3319, Veronica.Lance@noaa.gov

STAR Science Seminars

Presenter:
Dr. Benjamin T. Johnson - JCSDA

Sponsor:
STAR Science Seminar Series

Remote Access:
WebEx
Event Number: 905 776 033  
Password: STARSeminar

Event address for attendees:https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=md8d79cffbe41f5b7949986d2b669c0ed

Audio:  
USA participants: 866-832-9297
Passcode:  6070416

Abstract:
The Community Radiative Transfer Model (CRTM) is a fast, 1-D radiative transfer model designed to simulate top-of-the-atmosphere radiances consistent with a wide variety of satellite based sensors. The CRTM was primarily developed by JCSDA-funded scientists with essential contributions from NOAA/STAR and NOAA/EMC scientists. The primary goal of CRTM is to provide fast, accurate satellite radiance simulations and associated Jacobian calculations under all weather and surface conditions. CRTM supports all current operational and many research passive sensors, covering wavelengths ranging from the visible through the microwave. The model has undergone substantial improvement and expansion, since the first version in 2004. The CRTM has been used in the NOAA/NCEP and U.S. Navy operational data assimilation systems and by many other JCSDA partners such as NOAA/NESDIS/STAR, NOAA/OAR, NASA/GMAO, Naval Research Laboratory, Air Force Weather, and within multiple university environments. Over the past 14 years, both external research groups and operational centers alike have made essential contributions to the continued development and growth of CRTM.
A major goal of the CRTM core team is to ensure that CRTM becomes a true community radiative transfer model for all users. The CRTM official baseline code is developed and maintained based on internal and community-wide inputs, consisting of both improvements and externally contributed codes.
This presentation will briefly review the scientific and technical basis of CRTM, including its many strengths and limitations. There will also be an overview of the current status of the recently released CRTM version 2.3.0; and the future planned release of CRTM version 3.0.0 - which will represent a major milestone in CRTM's development and capabilities.

About the Speaker:
Dr. Benjamin T. Johnson joined NOAA/NESDIS/STAR (via AER, Inc.) in support of JCSDA in July 2015.  In January 2017, he was hired through UCAR as the JCSDA project lead for the Community Radiative Transfer Model (CRTM).    Dr. Johnson's primary responsibilities are to ensure that the CRTM project continues to be proactively developed and managed to meet operational user requirements.   This involves coordinating efforts and support for a large number of users and developers across a wide range of agencies and universities, both domestic and international.  
Dr. Johnson received a B.S. in Physics from Oklahoma State University, with an emphasis on hard-sphere sedimentation crystallization and photonics.  Combining his interest in weather, computing, and physics, he studied Atmospheric Science at Purdue University, where he received a M.S. degree. The next stop was the University of Wisconsin, where he completed his Ph.D. in Atmospheric Science advised by Dr. Grant Petty.
Before completing his Ph.D. in 2007, Dr. Johnson started working at NASA Goddard Space Flight Center in 2004 on the Global Precipitation Measurement mission, primarily focused on precipitation retrieval algorithm development and satellite observation simulations. During the intervening years, he has coordinated multiple NASA field campaigns as a mission scientist, and actively participates in the CGMS/WMO International Precipitation Working Group (IPWG), International TOVs Working Group (ITWC), and the International Workshop on Space-based Snowfall Measurement (IWSSM).   He is a member of the American Geophysical Union (AGU), and the American Meteorological Society (AMS).  
Dr. Johnson's primary areas of expertise are measuring and simulating cloud microphysical processes, theoretical and applied atmospheric radiative transfer, satellite remote sensing of clouds and precipitation, and satellite-based radar simulations in cold-cloud precipitating scenes.

STAR Science Seminars
This talk was originally scheduled for 11/15/2018 and 12/5/2018

Presenter:
Scott Rudlosky  - NESDIS/STAR/CoRP

Sponsor:
STAR Science Seminar Series

Remote Access:
WebEx:
Event Number:    903 126 314
Password: STARSeminar

Event address for attendees:
https://noaa-nesdis-star.webex.com/noaa-nesdis-star/j.php?MTID=m58ef2c99eed604b36781eae3ac7eb228

Audio:  
USA participants: 866-832-9297
Passcode:  6070416


Download slides:
https://www.star.nesdis.noaa.gov/star/documents/seminardocs/2018/20181115_Rudlosky.pdf

https://www.star.nesdis.noaa.gov/star/documents/seminardocs/2018/20181115_Rudlosky.pptx

Abstract:
The Geostationary Lightning Mapper (GLM) is the first sensor of its kind, and this technological advancement now allows continuous operational monitoring of lightning on time and space scales never before available. This has led to a golden age of lightning observations, which will spur more rapid progress toward synthesis of these observations with other meteorological datasets and forecasting tools. This study documents the first nine months of GLM observations, illustrating that the GLM captures similar spatial patterns of lightning occurrence to many previous studies. The present study shows that GLM flashes are less common over the oceans, but that the oceanic flashes are larger, brighter, and last longer than flashes over land. The GLM characteristics also help diagnose and document data quality artifacts that diminish in time with tuning of the instrument and filters. The GLM presents profound possibilities, with countless new applications anticipated over the coming decades. The baseline values reported herein aim to guide the early development and application of the GLM observations.

About the Speaker:

Dr. Scott Rudlosky is a NOAA/NESDIS physical scientist in the Center for Satellite Applications and Research (STAR) Cooperative Research Program (CoRP). He is co-located with the Cooperative Institute for Climate and Satellites (CICS) in College Park, Maryland. Scott is the NESDIS Subject Matter Expert on lightning and science lead for the Geostationary Lightning Mapper (GLM). He originally joined CICS as a Research Associate in January 2011 following completion of his M.S. (2007) and Ph.D. (2011) in Meteorology at Florida State University. He obtained his B.S. (2004) in Geography with a specialization in Atmospheric Science from Ohio State University.

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