Enter Note Done
Go to previous page in this tab
Session
  • Presentation | H32G: New Developments and Future Directions in Community Water Resources Modeling: Synergy at the Interface of Process Understanding, Artificial Intelligence, Computer Science, Operations, and Decision-Making II Oral
  • Oral
  • Share Icon
    Map Icon
    Bookmark Icon
  • H32G-05: Advancements in Coastal Community Modeling on the Cloud: Development and Implementation of the National Water Model Total Water Level Retrospective Run
    • Schedule
    Notes
  • Location Icon225-227
    NOLA CC
    Set Timezone
    •  
    View Map

Generic 'disconnected' Message
Author(s):
Trey Flowers, Office of Water Prediction, National Water Center (First Author)
Jason Ducker, NOAA Affiliate (Presenting Author)
Zachary Willis, NOAA Affiliate, Lynker
Mykel Alvis, Lynker
Katherine Moore Powell, NOAA Affiliate, Lynker, Office of Water Prediction, National Water Center, Tuscaloosa, AL, United States
Minna Ho, University of California Los Angeles
Michael Lalime, NOAA Affiliate, ISS
Tiffany Vance, U.S. Integrated Ocean Observing System
Patrick Tripp, RPS Oceans – USA
Micah Wengren, U.S. Integrated Ocean Observing System
Breanna Vanderplow, NOAA Affiliate, ISS, Inc.
Brian Cosgrove, NOAA/NWS/OHD
Camaron George, National Water Center, Office of Water Prediction, National Weather Service, NOAA
Joseph Zhang, Virginia Institute of Marine Science


The National Water Model version 3 (NWMv3) Total Water Level (TWL) capability predicts coastal flooding. Running these predictions for long periods, called hindcasts, are usually expensive and time-consuming using traditional supercomputers.


However, a new infrastructure called the Coastal Modeling Cloud Sandbox, developed by NOAA IOOS and partners, offers a cheaper way to do these long-term simulations using cloud computing. This infrastructure allowed us to develop and execute a 44-year NWMv3 TWL hindcast simulation in order to understand coastal flooding trends better.


This presentation will explain the NWMv3 TWL hindcast runs were constructed and executed in the Cloud Sandbox, including how we optimized the coastal modeling component to run efficiently and cost-effectively across cloud resources. Various forcing products from atmosphere and ocean models were pre-processed and post-processed using cloud resources and Zarr version3 data formatting for optimal performance and data storage. This work paves the way for future long-term coastal flooding predictions that are both affordable and efficient.




Scientific Discipline
Suggested Itineraries
Neighborhood
Type
Main Session
Discussion