skip to main content

Search for: All records

Creators/Authors contains: "Smith, Travis"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Abstract NOAA’s Hazardous Weather Testbed (HWT) is a physical space and research framework to foster collaboration and evaluate emerging tools, technology, and products for NWS operations. The HWT’s Experimental Warning Program (EWP) focuses on research, technology, and communication that may improve severe and hazardous weather warnings and societal response. The EWP was established with three fundamental hypotheses: 1) collaboration with operational meteorologists increases the speed of the transition process and rate of adoption of beneficial applications and technology, 2) the transition of knowledge between research and operations benefits both the research and operational communities, and 3) including end users in experiments generates outcomes that are more reliable and useful for society. The EWP is designed to mimic the operations of any NWS Forecast Office, providing the opportunity for experiments to leverage live and archived severe weather activity anywhere in the United States. During the first decade of activity in the EWP, 15 experiments covered topics including new radar and satellite applications, storm-scale numerical models and data assimilation, total lightning use in severe weather forecasting, and multiple social science and end-user topics. The experiments range from exploratory and conceptual research to more controlled experimental design to establish statistical patterns and causalmore »relationships. The EWP brought more than 400 NWS forecasters, 60 emergency managers, and 30 broadcast meteorologists to the HWT to participate in live demonstrations, archive events, and data-denial experiments influencing today’s operational warning environment and shaping the future of warning research, technology, and communication for years to come.« less
  2. Free, publicly-accessible full text available June 13, 2023
  3. This paper synthesizes multiple methods for machine learning (ML) model interpretation and visualization (MIV) focusing on meteorological applications. ML has recently exploded in popularity in many fields, including meteorology. Although ML has been successful in meteorology, it has not been as widely accepted, primarily due to the perception that ML models are “black boxes,” meaning the ML methods are thought to take inputs and provide outputs but not to yield physically interpretable information to the user. This paper introduces and demonstrates multiple MIV techniques for both traditional ML and deep learning, to enable meteorologists to understand what ML models have learned. We discuss permutation-based predictor importance, forward and backward selection, saliency maps, class-activation maps, backward optimization, and novelty detection. We apply these methods at multiple spatiotemporal scales to tornado, hail, winter precipitation type, and convective-storm mode. By analyzing such a wide variety of applications, we intend for this work to demystify the black box of ML, offer insight in applying MIV techniques, and serve as a MIV toolbox for meteorologists and other physical scientists.