National Water Model

Utilizing NOAA’s National Water Model (NWM) Reanalysis data for past hurricane events in the Hampton Roads Region

StormSense, an award-winning smart cities regional project of Hampton Roads, has integrated NOAA’s National Water Model (NWM) 25-year retrospective simulation dataset for past hurricanes as well as forecast data with real-time water level sensors to review flood inundation perspectives that leverages streaming data. Such integrations with geo-spatial data have been made possible with OmniSci Cloud. This will enable our understanding of data with a data science perspective through simulated patterns in the future.

The National Water Model (NWM) is a hydrologic modeling framework that simulates observed and forecast streamflow over the entire continental United States (CONUS). This dataset was released in 2016 by National Oceanic and Atmospheric Administration’s (NOAA) Office of Water Prediction (OWP) and continuous forecast data is currently available.

Registry of Open Data on AWS provides a 25-year (January 1993 through December 2017) retrospective simulation using version 1.2 of the National Water Model, and a 26-year (January 1993 through December 2018) retrospective simulation using version 2.0 of the National Water Model. Further detail of use is provided on the link.

We reviewed data from 15 events namely Matthew 2016, Julia 2016, Hermine 2016, Joaquin 2015, Arthur 2014, Sandy 2012, Irene 2011, Ida 2008, Ernesto 2006, Gaston 2004, Isabel 2003, Floyd 1999, Earl 1998, Bonnie 1998 and Bertha 1996.

Data Exploration – The data is provided in .netcdf format from the data registry. We reviewed the data and created a stream segment dataset for the Hampton Roads region by clipping with the regional boundary. Data downloads from AWS Open Data Registry is time intensive, potentially very space intensive as well.

  • NWM Forecast Ranges
    • Short Range Forecast configuration cycles hourly and produces hourly deterministic forecasts of stream flow and hydrologic states out to 18 hours
    • Medium Range Member 1 extends out to 10 days while members 2-7 extend out to 8.5 days
    • The Long Range Forecast cycles four times per day (i.e. every 6 hours) and produces a daily 16-member 30-day ensemble forecast.
  • Streamflow Data (Counterminous United States)
    • ~18 MB per hour over the Counterminous United States
    • 157 GB for one year
    • 3 TB for 20 years
  • Clipped Data
    • 3,774 streams in Hampton Roads
    • 18 years of streamflow data for Hampton Roads is 13 GB
    • 750 MB when compressed using LZMA compression

Sample data provided here was created from NOAA’s NWM data to create the maps. Here is the dataset location –

Data Deluge: Utilizing NOAA and StormSense Data for Monitoring Hurricane Flood RiskThis topic was presented at OmniSci Converge 2019 Conference, the conference for Accelerated Analytics & Data Science at the Computer History Museum on October 23, 2019 in Mountain View, California.

This visualization was created using OmniSci software product and shows the hurricane and tropical storm names, stream flows around time periods when storms have passed. The map shows the streams in Virginia Beach by selecting a which passed within a 65-mile radius of City of Virginia Beach
This visualization was created using OmniSci software product and shows the events Hermine, Julia and Matthew. The graph on the right shows Hurricane Matthew as the event selected and the small peaks on the graph towards the left shows lower intensity events Hermine and Julia. Short Range predictions were used in this study. NWM version 1.2 data was used in the study.

We have continued interest in extending the capabilities nationwide through conceptual framework with educational institutions and cloud technology companies that will help cities throughout the US using historical as well as predictive data that is currently being presented by NOAA and include more observational data into the models. The relationship with water level sensors and velocities will be studied and non contact velocity measuring can provide valuable information in the coastal areas during future events.


Scroll to Top