Scenario_2A_Water_Level_Signell Progress

[rsignell-usgs]

Progress: 2A Water Level Signell

Successes

• Near-real time observations and forecast model datasets were discovered, accessed and used
• Datasets were discovered from both Regional Associations and National Backbone providers
• Catalog-driven search using OGC CSW allowed for dynamic updating as new datasets become available (or inactive)
• Data from both OPeNDAP-CF and SOS endpoints were discovered, accessed and used.
• Standards-based tools allowed extraction of both unstructured and structured grid model output without any model specific code, thanks to the British Met Office Iris package.
• Comparison of observed and modeled water levels led to new insights regarding vertical datum issues which are being used to improve the models
  

Easy Challenges: Identified and Fixed

• None of the NOAA CO-OPS forecast model results (GBOFS, CBOFS, LHOFS, LOOFS, LMOFS, LEOFS, LSOFS, NYOFS, SJROFS, DBOFS, TBOFS) were initially accessible. These are the operational National Backbone products that are targeting water levels in many important ports.
  THREDDS catalogs for these 11 regions nationwide were developed to aggregate the data, and after testing, these aggregation catalogs were registered, making them discoverable and accessible. Make sure these NOAA CO-OPS model aggregation catalogs get harvested #24

• The NOAA CDSL ESTOFS forecast model results were not initially accessible. This is the operational National Backbone product designed to be the next-generation unstructured grid surge model for the Hurricane Center. A script was developed that cuts chunks from each forecast and copies them to a special directory where they can then be joined to form a continuous aggregation, a workaround to the problem that the Forecast Model Run collection does not work for unstructured grid data. Get aggregation going for the operational inundation model ESTOFS #22
  

Longer-term Challenges Identified

• NOAA CO-OPS Forecast model results using unstructured grids could not be aggregated because the THREDDS Data Server Forecast Model Run Collection does not yet work for unstructured grid data. This will require Unidata to include ugrid as a common data model type and enable it to function as a FMRC featureCollection. Work with Unidata to get ugrid adopted as a common data model feature type #84
• The script to allow ESTOFS output to be aggregated depends on the ncks routine from NCO Tools, which is waiting for admin permission to be installed on CSDL machines. Temporarily it's being run in Woods Hole until CSDL can take it over. Get aggregation going for the operational inundation model ESTOFS #22
• Iris issues:
  • The British Met Office Iris package can return lon, lat, depth and time from unstructured grid model results, but does not return the connectivity array that would be desirable for more advanced interpolation that the simple nearest neighbor approach used in this system test. To accomplish this a new unstructured grid class and methods might need to be developed for Iris. Work with British Met Office to add ugrid to Iris #85
  • Iris required a hack to allow sea surface height to be extracted as cubes. This is being addressed by the Iris development team. Iris 1.7dev problem with elevation, depth when used in formula_terms #43, Iris 1.7dev problem with sigma coordinate with more than 1 dimension #44