Download presentation
Presentation is loading. Please wait.
Published byDennis Wells Modified over 9 years ago
1
1 Report: JAG/OCM Operational Community Modeling COPC Meeting AFWA, Omaha, NB 2 May, 2007
2
2 Outline Update: WRF Development Testbed Center (DTC) - WRF Code Management & User Support plan - Inter-organization agreements for DTC management & resourcing - Resource trajectory and personnel hiring - Major DTC projects in FY07 and implications for COPC Update: WRF Science Code & Infrastructure (follow up COPC AI 2006-1.5) Update: WRF and ESMF
3
3 Update on DTC Management Progress: WRF code management & user support plan - MMM and DTC have reached agreement for NCAR to provide UNIFIED code management & support to WRF users. - NCAR has announced its first JOINT WRF TUTORIAL to be held summer 2007. - Both dynamical cores & multiple physics suites - New capabilities to be added to the tutorial as they become available – HWRF, HYCOM, ensembles, etc. Progress: Inter-organization agreements for DTC management & resourcing. - DTC is submitting proposal to NOAA for an NCAR-NOAA cooperative agreement. (Responds to NOAA ’07 BAA.) - DTC interagency agreement has been discussed between Air Force and NOAA for single transfer of all govt. money to DTC. - Can be expanded to include other agencies. - Potential DTC resources from NSF can follow direct path to NCAR. 3
4
4 NCAR Unified Code Management & User Support Plan Community Codes Developer Codes User Support: Documentation Tutorials Wrfhelp User workshops Code Developer Code Developer Code Developer Code Developers Committee Contributed Code Reference Code Operational Codes DTC testing and acceptance The WRF Code Repository Code Developer Courtesy NCAR DTC and MMM will share management and support of WRF system codes residing in a single WRF Code Repository. Accountability to development community for code entrance into WRF. Accountability to DTC Advisory Board for management and support of WRF Reference & Oper. codes. (Accountability for WRF software framework is TBD.)
5
5 Projected WRF DTC Resources: FY07-FY12 FY07FY08FY09FY10FY11FY12 AFWA400K 400.0K NOAA1,200K2,000K3,000K NCAR**250K FAA40K GSD* **300K NSF50K TOTAL2.24M3.04M4.04M *GSD “in-kind” scientist support. **Does not include computational resources and office space.
6
6 Expansion of DTC Personnel in FY07 Infrastructure & Reference code support – 2 software engineers hired. Data assimilation support – 1+ support positions announced for DA, including GSI. Verification system development – 3 temporary FTEs hired. - At least 1 FTE expected to develop as full-time support position. - AFWA is lead funder. HWRF support – 1 position to be announced for support of HYCOM and moveable two-way grid nesting. NMM dynamical core support – 1 assoc. scientist hired.
7
7 Major DTC Projects in FY07 Extension of WRF system capabilities: - Unified WRF verification system under development, building on existing codes: -NCEP forecast verification system (FVS) -GSD real-time verification system (RTVS) -NCAR object-oriented verification system for research - Unified WRF post-processor has been completed. Testing and evaluation of WRF to establish Reference code: - Dynamical core intercomparisons: Dx=13 km, fcst. 60-84 h - Purpose: Establish that both cores yield forecasts of similar accuracy, facilitating rapid R2O transfers of science and technology using the interoperable WRF system. - Purpose: Facilitate ensembles in operations and research. - Status: Experiment design is well advanced; Runs likely will start in summer ‘07. - Supports WRF goal of accelerating transfer of new science and technologies from research community. 7
8
8 Update on WRF Science Code and Infrastructure Development - 1 Coupling WRF to HYCOM – Development underway - NCEP/EMC: Preliminary work to couple HYCOM to WRF-NMM. - NCAR/MMM: HYCOM operating under the WRF infrastructure, using MCEL coupler. Preliminary testing underway. - Neither initiative currently attempts to perform coupling via ESMF. ESMF needs further regridding capabilities. WRFv2.2 released by NCAR/MMM (Dec. 2006) - Adds new microphysics & radiation parameterization (both cores). - Includes nudging FDDA (grid and obs), urban canopy model in Noah LSM, positive definite advection (ARW only). - Includes one-way static nesting code for NMM. - Includes new WRF Preprocessing System (WPS) for ARW core. - Next release: Will include WPS capability for NMM core.
9
9 Update on WRF Science Code and Infrastructure Development - 2 WRF Reference Code and Interoperability by DTC - Standardized regression testing under development for NMM nesting. - Developing new WRF Testing Framework (WTF) to improve future regression testing for Contributed and Reference codes. - Community requirements being collected. - 6 new physics combinations interoperable with NMM core in v2.2. - Testing more physics combinations for interoperability with NMM. - WRF Domain Wizard - GUI interface to define domains & run WPS. WRF Infrastructure, Streamlining and Convergence w/ESMF - ARW fully interoperable as an ESMF component (WRFv2.2). - ARW benchmark & HPC tests: 7 Teraflops on 12,500 processors.
10
10 Frameworks: Update on WRF and ESMF WRF software framework has undergone important advancements since Oct. ’06. - Full interoperability of WRF as an ESMF component - Capacity to run HYCOM under WRF framework NCAR/MMM in agreement with conclusion of the 2006 WRF-ESMF Convergence Workshop: “We believe [two] options are appropriate near term strategies. - WRF uses ESMF for high-level coupling, [and] - WRF uses ESMF low-level utilities.” MMM accountability for WRF software framework remains an unresolved element in code management plan. Versions of Navy and NCEP models have been/are being developed to run as ESMF components. - GFS and NMM at NCEP - NOGAPS and COAMPS at NRL 10
11
11 Future Directions of WRF and ESMF - 1 WRF software framework is reliable, reasonably mature and will remain the dominant framework for community mesoscale modeling for at least 5-8 years. NCEP has offered to make its ESMF-NMM available to some external users for testing and evaluation, as soon as feasible. (probably in late ’07 or early ’08) Goals: (1) - Assist NCEP with model testing. (2) - Evaluate ESMF as a potential framework for future mesoscale modeling in the research community. NCEP expects to have its regional and global models converted to an ESMF-based system in ~5 years. Navy modeling systems also anticipated to migrate to ESMF on approximately same time scale. 11
12
12 Future Directions of WRF and ESMF - 2 Maintaining benefits of accelerated transfer of new modeling science and technologies between research and operations will require a common modeling framework. DTC has resources to become a logical focal point for extending ESMF-based models into an interoperable, multi-core, multi-physics community modeling system supported to users. - Too early to estimate probability of success. - Warrants ongoing development and testing. - Time horizon – 6-10 years to fully transition the community modeling system from WRF to ESMF.
13
13 Recommendations on WRF and ESMF Monitor ongoing development of ESMF as it adds range of capabilities required for mesoscale applications. Monitor deliberations of NUOPC. Continue development, testing and evaluation of ESMF-based limited-area models at NCEP and NRL. Encourage broad community involvement in testing and evaluation ESMF-based models from NRL and NCEP. Identify extensions and developments necessary for ESMF and models running under ESMF to be considered as a potential community-supported system. Assess time and resources required to follow development and implementation path to a community system. COPC has the potential to contribute significantly to all the above activities w/o derailing community modeling.
14
14 End
15
15 FY07 - WRF Development Testbed Center: A Growing Resource Supporting WRF & COPC AFWANOAANCARFAAGSDTotal 1. Software Management151.5K50.5K0.0K 202.0K 2. Support User Comm.168.5K392.5K199.5K0.0K 760.5K 3. T&E of Contrib. Code0.0K161.0K0.0K41.0K202.0K404.0K 4. T&E of Refer./Oper. Codes 0.0K154.0K0.0K 48.0K202.0K 5. Transition to Oper.0.0K202.0K0.0K 202.0K 6. Administration80.0K240.0K50.5K9.0K0.0K*379.5K TOTAL COST400.0K1200.0K250.0K50.0K250.0K2150.0K *Since GSD money shown here is “in-kind” scientist support, no administrative overhead is charged by DTC. Table 1. Proposed allocation of DTC FY07 funding by agency, (as of Oct. 2007)
16
16 WRF Development Testbed Center: A Growing Resource Supporting WRF & COPC Proposed DTC functions in FY07: WRF software version control and definition of Reference Codes Software framework management & development prioritization Reference code support and documentation Ensemble modeling & post-processing support Verification system development and support Visiting scientist program Establishment and maintenance of physics interoperability Testing for elevation of contributed codes to Reference status NWP science support Testing and evaluation of Reference codes Transition of specific candidate codes to operations centers
17
17 CSAB AI-2006-5.6: Proposed Streamlining Tasks for WRF Software Framework 1. Develop and publish detailed documentation of the WRF I/O stack (API). Cost: ~$22K, Time: ~Spring ‘07 2. Confirm the reduction of the WRF Registry due to elimination of obsolete sections.Current phase completed; More possible in ’07. (No funding requested.) 3. Standardize initialization of physics packages. Cost: $130K (NCAR&NRL), Time: 1 yr? 4. Develop and publish standardized build mechanisms and conventions. Cost: $90K (NCAR&NRL), Time: ~1 yr 5. Develop and publish standards for parameter naming & metadata conventions.Cost: none requested, Time: depends on Clim.Fc. 6. Adopt climate forecast metadata conventions for geo-registration data. Cost: none requested, Time: depends on Clim.Fc. 7. Establish metadata conventions for generic model info (start time, dt, etc.) Cost: none requested, Time: depends on Clim.Fc. 8. Provide interoperability source-code “translator” for physics using different k-index and looping orders.Cost: ~$275K for prelim. phase, Time: ~1-2 yr 9.Reassess comparative efficiency of WRF codes in (i,j,k) vs. (i,k,j). MMM does not consider this task a high priority.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.