1. A Super-Regional Modeling Testbed for Improving Forecasts of Environmental Processes for the U.S. Atlantic and Gulf of Mexico Coasts Don Wright, SURA Principal Investigator Rich Signell, USGS Technical Advisory and Evaluation Group Chair

2. Motivation for the testbed: improving prediction of environmental processes Design of this testbed Year 1 products Future work Outline

3. Improving Forecasts of Coastal Environmental Processes Factors: open boundary conditions, surface and river forcing conditions, enhanced physics, adjustable parameters, data assimilation, numerics, amount of data assimilated, skill of modelers(!), vertical and horizontal resolution, coupling to wave and met models. “Which model is better?” is not the right question. What factors in the simulation resulted in a better solution? How much better? At what cost?

4. Defining Improvement To measure improvement for environmental processes, we need to define skill metrics for specific environmental processes and often for specific region Inundation, search and rescue, deep oil spills, navigation, hypoxia, harmful algal blooms, diver operations, alternative energy siting, beach erosion, regional impact of climate change all require different skill metrics Operational centers need community help in this process – too broad for NOAA and NAVY!

5. The ocean community needs a common cyberinfrastructure to access, analyze and display data from the different models: each model currently has their own standards and toolsets A Common Cyberinfrastructure for Model Data Structured Grids Unstructured Grid 5x5 6x3 10 nodes Variety of Stretched Vertical Coordinates

6. The Plan: Build a common infrastructure to enable access, analysis and visualization of all coastal ocean model data produced by NOAA, NAVY and IOOS Develop skill metrics and assess models in three different regions and dynamical regimes, to ensure a robust and powerful infrastructure Identify factors that could be transitioned to operations Build stronger relationships between academia and operational centers through collaboration A Testbed Framework for Coastal Ocean Models

7. Testbed “Management” Testbed Advisory Evaluation Group Shelf Hypoxia Gulf of Mexico Estuarine Hypoxia Chesapeake Bay Cyber Infrastructure Coastal Inundation Gulf and East Coast Testbed Team Structure Rick Luettich, UNC-CHJohn Harding, MSU Carl Friedrichs, VIMS Rich Signell, USGS Eoin Howlett, ASA Don Wright, SURA Doug Levin, NOAA/IOOS Liz Smith, SURA 25 members 21 members 20 members 24 members 7 members

8. Cyberinfrastructure (CI) All Regions – All Teams  Extending CI from OGC, Unidata and others (NOAA DMIT, USGS CDI) to support unstructured grids, and add functionality  Web Access via OpenDAP w/CF  Unidata Common Data Model/NetCDF Java Library API  Distributed search capability  Browser based map viewer (WMS)  Toolbox for scientific desktop analysis  All components standards-based! Search services Mapping services and browse application Analyze in scientific desktop application

9. Inundation Extra-tropical – Gulf of Maine Tropical – Gulf of Mexico - 4 models: 3 unstructured grid +1 structured grid - Coupled wave-storm surge-inundation (TWL) - Consistent forcing, validation and skill assessment using existing IMEDS tool -Extensive observational data sets for historical storms Ike, Rita and Gustav in standard formats -SURA has provided supercomputer resources Inundation Extra-tropical – Gulf of Maine Tropical – Gulf of Mexico - 4 models: 3 unstructured grid +1 structured grid - Coupled wave-storm surge-inundation (TWL) - Consistent forcing, validation and skill assessment using existing IMEDS tool -Extensive observational data sets for historical storms Ike, Rita and Gustav in standard formats -SURA has provided supercomputer resources Extratropical Grid Tropical Grids for Galveston Bay

10. Estuarine Hypoxia Chesapeake Bay 1. Estuary: – 5 Hydrodynamic models – 3 Biological (DO) models – 2004 data from 28 CBP stations – Comparing T, S, max (dS/dz), DO via target diagrams 2. Shelf: OBCs 5 hydrodynamic models Estuarine Hypoxia Chesapeake Bay 1. Estuary: – 5 Hydrodynamic models – 3 Biological (DO) models – 2004 data from 28 CBP stations – Comparing T, S, max (dS/dz), DO via target diagrams 2. Shelf: OBCs 5 hydrodynamic models Models doing better on oxygen than stratification! Stratification (dS/Dz) Dissolved Oxygen

11. Shelf Hypoxia Gulf of Mexico Hydrodynamic & biogeochemical hindcast comparisons of hypoxia model (stand alone) coupled to 3 different Gulf of Mexico hydrodynamics models Evaluation of two shelf hypoxia formulations (NOAA & EPA) Shelf Hypoxia Gulf of Mexico Hydrodynamic & biogeochemical hindcast comparisons of hypoxia model (stand alone) coupled to 3 different Gulf of Mexico hydrodynamics models Evaluation of two shelf hypoxia formulations (NOAA & EPA)

12. Foundation of a cyberinfrastructure framework for search, access and display of all NOAA, NAVY and IOOS model data, via browser and scientific desktop application Skill metrics and identification of key performance factors and cost for three important dynamical regimes and environmental issues CONOPS for transition from research to operations Improved communication between research and operations Testbed Year 1 Products

13. Expand to more regions and problems Examine more factors (e.g. data assimilation) Build out the cyberinfrastructure Conduct training in the community Sustaining future development Future Work for the Testbed (1/2)

14. “How much do I get for how much?” is a better question than: “How much does a testbed cost?” $250K full time developer to help people get their data connected, maintain the servers, build (limited functionality into the toolkit) $500K for three investigators to look at a particular issue, develop metrics, compare models (deep oil spill, harmful algal bloom, different data assimilation techniques) $4000K for another year like Year 1 Future Work for the Testbed (2/2)