Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 FIRST ANNUAL NUOPC WORKSHOP Preliminary Report Scott Sandgathe NUOPC Technical Lead 19 Aug 2010 Scott Sandgathe NUOPC Technical Lead 19 Aug 2010.

Similar presentations


Presentation on theme: "1 FIRST ANNUAL NUOPC WORKSHOP Preliminary Report Scott Sandgathe NUOPC Technical Lead 19 Aug 2010 Scott Sandgathe NUOPC Technical Lead 19 Aug 2010."— Presentation transcript:

1 1 FIRST ANNUAL NUOPC WORKSHOP Preliminary Report Scott Sandgathe NUOPC Technical Lead 19 Aug 2010 Scott Sandgathe NUOPC Technical Lead 19 Aug 2010

2 Outline NUOPC Vision and Goals NUOPC Progress Report NUOPC R&D Workshop oQuick look! NUOPC Vision and Goals NUOPC Progress Report NUOPC R&D Workshop oQuick look! 2

3 3 NUOPC STRATEGIC VISION A National System with a Tri-Agency commitment to address common requirements oTechnical architecture oCommitting sufficient funding, facilities and human capital resources oFocusing research on coordinated national objectives Multi-component system with interoperable components built upon common standards and a framework such as the ESMF

4 4 Managed ensemble diversity Significantly improve forecast accuracy Quantify, bound, and reduce forecast uncertainty Joint ensemble Produce most probable forecast, e.g. high impact weather Provide mission specific ensemble products Drive high-resolution regional/local predictions Drive other down stream models. Establish a national global NWP research agenda to accelerate development and transition Managed ensemble diversity Significantly improve forecast accuracy Quantify, bound, and reduce forecast uncertainty Joint ensemble Produce most probable forecast, e.g. high impact weather Provide mission specific ensemble products Drive high-resolution regional/local predictions Drive other down stream models. Establish a national global NWP research agenda to accelerate development and transition NUOPC STRATEGIC VISION (2)

5 5 GOALS Accelerate improvement of our National capability by: oImplementing a global atmospheric ensemble system designed to enhance predictive capability oClearly articulating operational requirements and a corresponding National research agenda, with initial emphasis on hurricane track/intensity forecasts, joint wind and seas forecasts, and ceiling/visibility forecasts oSharing the predictive burden among the operational agencies oPromoting collaboration on development among government agencies oAccelerating the transition of new technology into the operational centers oImplementing ways to enhance broad community participation in addressing the National research agenda

6 6 NUOPC PROGRESS REPORT Implementation of an operational multi-agency global atmosphere ensemble oInitial capability based on NAEFS scheduled for 16 Dec 2010 oCurrently successfully passing fields from FNMOC to NCEP for testing in NAEFS ensemble Agreement on how to manage a multi-agency ensemble in an operational environment oBeing addressed by Committee for Operational Processing Centers (COPC) oAnnex to current Data Acquisition, Processing, and Exchange (DAPE) agreement Shared post processing agreement oWorking group addressing who, where, when, how much oDevelopment of common post processing toolbox

7 77 COMMITTEE PROGRESS REPORT Committees currently addressing details of greater collaboration and coordination oCommon Model Architecture Committee addressing overall architecture and future standards (presentation later) oContent Standards Committee working with Earth System Modeling Framework (ESMF) committees to address standardized implementation of the Earth System Modeling Framework and development of a NUOPC layer. oTechnology Transition Processes Committee addressing common operational needs, developing a research and development agenda and outreach including hosting this workshop. oMetrics Subcommittee working on agreement for standard evaluation of model and ensemble performance, standard climatology, standard test cases, metrics for operational impact

8 8 WORKSHOP GOALS Bring the operational community together with the research and development community and the funding agencies to arrive at a requirements-driven, prioritized research agenda for global atmospheric ensemble prediction and post processing.

9 9 OBJECTIVES Review NUOPC progress Review Agency operational needs and identify common needs Review ongoing ensemble research Propose prioritized research and development agenda to meet common needs Review NUOPC progress Review Agency operational needs and identify common needs Review ongoing ensemble research Propose prioritized research and development agenda to meet common needs

10 10 Session 3: OPS Needs vs Ensembles Consolidated Report?? Given the operational needs presented: What is common? Which ones are appropriate for ensembles to address? Where do ensembles have to be in 10 years to meet these needs? What infrastructure is required?

11 Common needs Better support for decision making oMission-related (protection of assets, safety of personnel – Navy and AF) oEmergency management (life and property – NOAA) Common interests oAviation (turbulence, icing, ceiling, visibility) oTropical cyclones track and intensity oMaritime safety and routing oAir quality and visibility oSurface weather variables oHigh impact events

12 Common needs (2) Improved Global Forecast Performance Support Coupling (mesoscale, ocean, etc) Sharp and reliable probability based products/tools to support improved decision making Improved deterministic sensible weather forecasts. Tools for interpretation of ensemble output for improved decision making (More?) Training of forecasters and other end users in the use of ensemble products, including social scientists. (More?)

13 Common needs (3) Improved means for gathering user requirements and understanding decision processes. Improved verification metrics and scorecards, including extension into end user decision making processes and the production and archiving of an “analysis of record. Post Processing Toolkit (Calibration) Communication/Visualization Tools oGlobal and Mesoscale Training, Education, and Marketing Metrics Research to Ops Support Common Standards and Architecture Research support tools (data, archive, etc)

14 Infrastructure required The number one mentioned infrastructure requirement by all groups was computational resources adequate to handle high resolution EnDA, ensembles, and post processing including reforecasting, to support long term research. All groups felt adequate communications were required to pass, access rapidly, and distribute complex ensemble fields and products. All groups mentioned tools and software support for efficient data access to enhance R2O. Two groups mentioned data storage adequate to handle very large historical ensemble data sets and to support long term research. Two groups mentioned training or training infrastructure. Two groups mentioned outreach or marketing with the assistance of social scientists – infrastructure to support customer feedback. One group mentioned verification. One group mentioned adequate observations – infrastructure to assess data needed to meet goals.

15 Big Picture - Needs Computer Resources, infrastructure oFor operations oFor transition to ops oFor research on operational systems Marketing infrastructure oAssess user needs oAssess user value oPresent/adapt new products Computer Resources, infrastructure oFor operations oFor transition to ops oFor research on operational systems Marketing infrastructure oAssess user needs oAssess user value oPresent/adapt new products 15

16 16 Needs driven research agenda Given the common needs for ensembles: What are realistic goals? What are the critical research areas? What are the gaps? Can we prioritize? What specific additional resources are required?

17 17 Realistic Goals Two day forecast improvement/decade in global scale Improve the prediction of position and timing of mesoscale phenomena to match nowcasting. Reduce sensible prediction errors from 6 hr to 1 hr in below 20 km scale. Nonhydrostatic, new dynamic core global forecast system. Fully coupled earth prediction system, air, ocean, sea ice, land, and wave models Coupled data assimilation

18 18 Realistic Goals Research to inform what makes sense on how structure operations. Improved estimation and quantification of weather related socio-economic impacts Continue to drive model predictability (and forecastability) towards true predictability Close the Valley of Death (testbeds, application research, tools/applications)

19 19 Critical Research Areas Trade off between number of ensemble and resolutions. Initial condition uncertainty oCritical for small scales and multi-scale phenomena. oHow to perturb the initial conditions. oObs uncertainty vs background uncertainty. Improve the initial condition of the deterministic model. Model uncertainty. oDynamics and physics

20 20 Critical Research Areas Multi-scale error growth. Verification and calibration for ensemble forecast, oPost processing, oTools for customers oRe-forecasts Improved understanding of atmospheric processes and phenomena The science of prediction oReducing and accounting for model inadequacy oData Assimilation oObserving systems oIntrinsic midlatitude predictability oThe error dynamics of ensemble prediction systems

21 21 Critical Research Areas Decision and Behavior research, oTies with the funding opportunity. oEconomic value/benefit. oOperational Risk management tool. oDecision based on cost savings. Best communication tools. Decision science research and applications oUse of forecast information in decision making oCommunicating forecast uncertainty oUser-relevant verification oQuantifying the value of forecasts oDeveloping decision support tools and systems

22 22 Critical Research Areas Analysis related uncertainties: oBest way to exploit ensemble covariance info 2-way coupling oDealing w/ non-Gaussian error distributions Particularly for fine scale Moisture oNon-linearity and non-locality in observation operators oCoupled system oData preprocessing QC Thinning oTools for assessing observing systems

23 23 Critical Research Areas Uncertainty exploitation, customer dependent. Model related uncertainties: –Model physics paradigm shift Cloud resolving model Probabilistic model physics parameterization –Multi-model vs. one model w/ stochastic physics/parameters/etc. Configurations for multi-model Acquisition lifecycle cost considerations

24 24 Critical Research Areas –Calibration Best technique(s) (w.r.t. applications, ops) Min training datasets (variables, scales, etc.) Techniques for high-dim problems (e.g. aircraft route) Handling derived variables Train on observations or analyses Modular (e.g., bias cor. first) vs. holistic approach Re-forecasts

25 –Applications, Forecast System Configuration Multiscale modeling (global, regional, or mixed), value to customer, and costs in ops Best way to interact with users – feedback loop into products Ways to effectively educate Propagating uncertainty into expensive downstream applications (storm surge, dispersion, etc.) Critical Research Areas

26 Verification: oProbabilistic Metrics for high-dimensional, multi-variate applications oMORE data, and data mining tools HPC design oSoftware engineering oGPU? oExploit new architectures oFault tolerance Verification: oProbabilistic Metrics for high-dimensional, multi-variate applications oMORE data, and data mining tools HPC design oSoftware engineering oGPU? oExploit new architectures oFault tolerance Critical Research Areas

27 Big Picture – R&D Ask me tomorrow?? 27


Download ppt "1 FIRST ANNUAL NUOPC WORKSHOP Preliminary Report Scott Sandgathe NUOPC Technical Lead 19 Aug 2010 Scott Sandgathe NUOPC Technical Lead 19 Aug 2010."

Similar presentations


Ads by Google