1 / 26

Joint Action Group for Operational Community Modeling

Joint Action Group for Operational Community Modeling. Fall COPC 2007. Dr. Nelson Seaman JAG/OCM Chair. Overview. Membership Major Accomplishments - DTC Technical Review Pending Issues - DTC Management and Resourcing - WRF & ESMF. JAG/OCM Membership.

marly
Download Presentation

Joint Action Group for Operational Community Modeling

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Joint Action Group for Operational Community Modeling Fall COPC 2007 Dr. Nelson Seaman JAG/OCM Chair

  2. Overview • Membership • Major Accomplishments • - DTC Technical Review • Pending Issues • - DTC Management and Resourcing • - WRF & ESMF

  3. JAG/OCM Membership • NWS – Nelson Seaman (Chair, through Jan. 2008) • FNMOC – Roger Stocker • AFWA – Capt. Leonard Berman • NAVO – Frank Bub • NRL/MRY – Richard Hodur • NRL 7300 (Oceanog. Div., Stennis) – Steven Payne Action: Need COPC to appoint next JAG/OCM Chair. (Seaman’s term expires 3 Feb. 2008.)

  4. Major Accomplishment • Completed technical review of Developmental Testbed • Center (DTC) under auspices of WRF ExOB. • - Participants – Chief sponsors of DTC are: • - NOAA/NCEP • - NOAA/OAR • - AFWA • - NCAR • - Some key assessments & recommendations: • - DTC is accomplishing its mission for management, • support and testing of community codes. • - Replace concept of “Reference Code” • by more specific “Reference Configurations.” • - Expand visiting scientist program, as funding allows. • - Keep careful track of how agency funds are used.

  5. Issue 1: DTC Management and Resourcing • Management Plan • - Draft DTC Charter progressing through review • process. • - Review inputs received from all agencies. • - No obvious “show stoppers”. • - Next step: • - Integrate agency recommendations. • - Return to management for final reconciliations • and legal review prior to signing. • Resourcing • - DTC proposal for 5-year support under NOAA • BAA (similar to COMET) is under review.

  6. Issue 2: WRF and ESMF Options for Living with Two Frameworks

  7. Background: Recommendations to COPC, 2 May 2007 • Monitor ongoing development of ESMF. • Monitor deliberations of NUOPC. • Continue development and testing of ESMF-based NWP models. • Encourage community involvement in T&E of ESMF NWP models. • Identify extensions necessary for ESMF-based models to • meet broad NWP community requirements. • Assess time and resources required to develop an ESMF-based • community modeling system. These were mostly passive remedies.

  8. New Decisions Bring New Challenges • NOAA has adopted ESMF for its unified modeling systems. • NUOPC has agreed to adopt ESMF for future model applications. • Some other major users and many researchers want to • continue using WRF CMI. • ESMF may not replicate many desirable WRF functions • (e.g., nested grids) for ~4-8 years. • Lengthy lead times for critical-path decisions & code development • require a flexible solution that allows users to… • - run under ESMF (w/o WRF CMI), if desired. • - continue running under WRF CMI, if desired. • - maintain & support community code. • - interoperability, dynamics & physics options • - two-way transfer of S&T innovations • (R2O and O2R) Prime NUOPC requirement Prime NCAR/MMM requirement In everybody’s interest

  9. NEW DEVELOPMENT: ESRL/GSD Proposal for Rapid Refresh Model • Goal: • Extract ARW from WRF-CMI. • Create a parallel ARW version under NCAR’s ESMF for RR. • Basis: • Most WRF science codes for dynamics and physics are same or very similar under both frameworks! • NMM-B provides a working example (already in ESMF). • Method: • First, make a “by-hand” translation of ARW into ESMF. • Create an automated pre-processor translation code to minimize transfer effort for future ARW developments. • Cost Estimate: ~2.4 FTEs (modest) • Late-Breaking News: T. Henderson has moved from MMM to GSD. Courtesy – Stan Benjamin, ESRL/GSD

  10. Three Options for Common Path Forward • Option 1: Support one WRF-based repository, but • with parallel science codes under ESMF used at • OPCs and two-way translation of codes. • Option 2: Establish two separate code repositories • under WRF and ESMF, both supported to users. • (No significant NCAR/MMM participation.) • Option 3: Establish single ESMF repository; engage • NCAR/MMM participation to accelerate transfer • of most critical functions from WRF-CMI to ESMF.

  11. Option 1: Extend GSD’s Proposal to NMM-B Proposal: Develop a two-way translation capability (pre-processor) by extending the one-way translator concept of ESRL/GSD. WRF Software Framework: Supported to research comm. ESMF Framework: (Shown as configured for NCEP operations) Module_MAIN_GRID_COMP ARW solver solve_em.F NMM-B solver solve_nmm.F coupler dynamics module physics module WRF Physics Interfaces ARW-ESMF compatible solver NMM-B ESMF compatible solver Plug-Compatible Physics Plug-Compatible Physics Red dashed line denotes GSD’s ARW translation path; Blue dashed line denotes NMM translation path. Note – Diagrams above are highly simplified representations of fairly complex codes.

  12. Summary of Option 1 • Code Support:Single WRF CMI repository remains basis for • supporting shared code system to user community. • Advantages: • Protects agency investments in WRF CMI and its advantageous • flexibility/functionality desired by many users. • Provides two-way transition path for S&T advancements & • physics interoperability. • No impact to users who want to work… • - ONLY with ESMF, or • - ONLY with WRF CMI. • DTC runs translator codes once per new release of versions. • - Most users never touch translator codes. • Disadvantages: • Requires NMM-B installation into WRF CMI. (complex task?) • Cost - Initially high (source uncertain); moderate maintenance cost. • Risk - MMM again has just one support expert for WRF CMI.

  13. Option 2 : Move To Separate Repositories Proposal: Create a competingESMF shared repository supported to user community, based on GSD’s one-way translation concept. Global Space Wx Ocean Wave Land Sfc. Hydrolog. WRF Software Framework: Supported for research & ops ESMF Framework: Supported for research & operations frozen Module_MAIN_GRID_COMP ARW solver solve_em.F NMM-E solver solve_nmm.F coupler dynamics module physics module WRF Physics Interfaces ARW-ESMF compatible solver NMM-B ESMF compatible solver Plug-Compatible Physics Plug-Compatible Physics Red dashed line denotes GSD’s ARW translation path. Note – Diagrams above are highly simplified representations of fairly complex codes.

  14. Proposed Steps for Option 2 • 1. NCAR/MMM continues to support WRF-ARW and WRF repository. • DTC establishes/supports new ESMF-based community code • repository containing… • NCEP’s NMM-B and physics • GSD’s ESMF-ARW and NCAR physics • Tri-agencies become responsible to extend (or provide resources • for) ESMF-based model codes to include grid nesting, pre- and • post-processing, & other WRF functions needed by researchers. • MMM extends/supports WRF-ARW as global-regional system for • research. NMM-E becomes obsolete in WRF. • NCEP extends/supports ESMF-NMM-B as global regional system • for operations. DTC supports it for research use. One of the two repositories eventually may be shut down (frozen) if number of users fails to justify ongoing costs.

  15. Summary of Option 2 • Code Support: - NCAR/MMM supports WRF-CMI repository; • - DTC supports ESMF-based repository. • Advantages: • Protects agency investments in WRF CMI, at least until ESMF-based • system develops proven flexibility & functionality. • Maintains physics interoperability. • Closer to NUOPC’s agreed path and links to global research commun. • Little immediate impact to users who want to work… • - ONLY with ESMF, or • - ONLY with WRF CMI. • Disadvantages: • Cost – Long-term development cost is high; maintenance cost is high. • Risk – NMM-E becomes obsolete; WRF effectively a one-core system. • Risk – Divergent paths for operations & research NWP more likely. • Risk – If ESMF-based system may attract few research users. • Divergent capabilities – Support experts less able to share work load.

  16. Option 3 : Move To ESMF Repository Proposal: Create a new ESMF-based shared repository supported to user community; gradually decommission WRF repository. Global Space Wx Ocean Wave Land Sfc. Hydrolog. WRF Software Framework: Eventually decommissioned ESMF Framework: Supported for research & operations frozen Module_MAIN_GRID_COMP ARW solver solve_em.F NMM-E solver solve_nmm.F coupler dynamics module physics module WRF Physics Interfaces ARW-ESMF compatible solver NMM-B ESMF compatible solver Plug-Compatible Physics Plug-Compatible Physics Red dashed line denotes GSD’s ARW translation path. Note – Diagrams above are highly simplified representations of fairly complex codes.

  17. Proposed Steps for Option 3 • 1. NCAR/MMM & DTC temporarily continue support for WRF repository. • 2. DTC & MMM establish new ESMF-based community repository. • NCEP’s NMM-B and physics • GSD’s ESMF-ARW and NCAR physics • Tri-agencies provide resourcesfor ESMF-based model codes • to add grid nesting, pre- and post-processing, & other WRF • functions needed by users, with MMM and NOAA as lead • developers. • MMM extends ARW as global-regional system under ESMF and • supports it to research community. • NCEP extends ESMF-NMM-B as global regional system for • operations. DTC supports it for research use. • MMM decommissions WRF-CMI repositorywhen ESMF system • meets critical user requirements.

  18. Summary of Option 3 • Code Support: - NCAR/MMM & DTC share support of common ESMF-based • code repository. • Advantages: • Protects agency investments in WRF CMI until ESMF-based system • develops proven flexibility & functionality. • Maintains physics interoperability, dynamics options, etc. • Consistent with NUOPC’s agreed path & links to global res. commun. • Maximizes leveraging of resources for system development and code • support/maintenance (funding and personnel). • Little immediate impact to users who want to work… • - ONLY with ESMF, or • - ONLY with WRF CMI. • Disadvantages: • Risk – Not clear that OPCs and NCAR can agree on this path or a • technical design.

  19. Final Cautionary Note • All aspects of these options must be evaluated by • managers and technical experts for… • - responsiveness to requirements • - technical feasibility • - cost & risk • GSI, supported through JCSDA, can be used with any of the options • described above.

  20. Questions?

  21. Backups

  22. Implications of NUOPC1 and ESMF2 for WRF Community Modeling (from WRF ExOB meeting - 31 July 2007) • Option: Create and maintain a separate code repository under ESMF, • chiefly for operations. Current WRF repository continues to exist. • Recommendation was:Reject separate repositories. • Risk: Two code repositories, one under ESMF and one under WRF CMI3, • would probably be difficult and costly to maintain. • - WRF code maintenance is already shared by DTC and NCAR/MMM, • although using a single repository. • - Two repositories maintained by two organizations is more complex. • - Community modeling partnership may become unsustainable. • Need practical way forward that… • 1. Does not impede progress for research or operations (no harm). • 2. Preserves single code repository as basis for community modeling. • 3. Contains no single points of failure. 1 – National Unified Operational Prediction Capability 3 – Common Modeling Infrastructure 2 – Earth System Modeling Framework

  23. Final Recommendation to WRF ExOB (31 July 2007) • ACTION ITEM 7.6-2007: • Appoint a management team to… • Identify chief functional requirements needed for a software framework • to support WRF science codes (from perspective of limited-area WRF users ). • Appoint a technical/managerial team to… • Determine status of each necessary WRF CMI function in ESMF. • - function exists now • - function is planned (timetable?) • - function is not currently planned • Analyze potential to develop (or accelerate development of) desired • functions needed for ESMF. • - maintain external “look and feel” of WRF for users. • - ensure user interaction is straightforward. • Develop cost and time estimates for all new functions. • …and report to ExOB. Implies only one software framework might survive, long term.

  24. Structural Relationships for Mapping of NCEP’s NMM Model Module: _DYNAMICS-Init-Run-Finalize.F (for NCEP’s ESMF-based version of NMM-B model) ESMF infrastructure, ESMF routines Subroutines called: Section 1 – Momentum / BC HDIFF BOCOH PGFORCE UPDATEUV BOCOV DHT DDAMP Section 2 – Continuity / advection PDTSDT ADV1 VTOA UPDATET CDZDT CDWDT VSOUND VADV2 Module:_PHYSICS-Init-Run-Finalize.F Section 3 – Physics called via physics drivers. Solve_nmm.F (for WRF-based version of NMM) WRF infrastructure, including use of Registry for variable definitions Subroutines called: Section 1 – Continuity / advection Most science subroutines in ESMF version are same or nearly same as in WRF. PDTE ADVE VTOA VADZ HADZ EPS VADV2 HAD2 Section 2- Physics called via physics drivers Section 3 – Momentum / BC HDIFF BOCOH PFDHT DDAMP BOCOV Courtesy – Stan Benjamin, ESRL/GSD

  25. Mapping Structure of ARW from WRF to ESMF ESRL/GSD Rapid Refresh Proposal – July 2007 One-way translation of ARW from WRF CMI into ESMF using NCEP’s two-component configuration Adding NMM can provide future option for RR ensemble WRF Software Framework ESMF Framework NMM-B solver (stand-alone NCEP code) Module_MAIN_GRID_COMP ARW solver solve_em.F NMM-E solver solve_nmm.F coupler WRF Physics Interfaces Dynamics module Physics module Plug-Compatible Physics Plug-Compatible Physics Cost estimate by GSD: ~2.4 FTEs Courtesy – Stan Benjamin, ESRL/GSD

  26. Summary: GSD’s One-Way ARW Translation • Advantages: • - ARW components are already CMI-compliant modules. • (“Straight-forward to repackage as ESMF-compliant modules.”) • - Physics codes remain same in WRF and ESMF. • - NCEP will not have to run WRF CMI for operational RR model. • - Future ARW developments can be transitioned readily into ESMF. • - Facilitates use of ARW in future ESMF-based applications at NCEP. • (SREF and HWRF) • Disadvantages: • - Cost is non-trivial, but probably acceptable. (Est’d. ~2.4 FTEs) • - Provides no O2R path into WRF CMI for future NMM-B developments. • (WRF contains NMM-E) • - Some important WRF functionality is lost. Note – GSD’s proposal is designed to address NCEP operational needs and is consistent with NUOPC agreement.

More Related