Download presentation
Presentation is loading. Please wait.
Published byBeverly Cox Modified over 9 years ago
1
1 NCEP Operational Prediction: Current status and future plans Stephen J. Lord Director NCEP Environmental Modeling Center NCEP: “where America’s climate and weather services begin”
2
2 Overview Mission comparison with International Centers –Current model suite Performance comparison –Global NWP –Hurricanes –S/I Climate –Precipitation (US) Improved strategy for forecast system enhancements Future model suite Summary
3
3 NCEP Mission Requirements & Forecast Suite Elements Suite Elements Global NWP Meso NWP Fire Wx Rapid Update Reg. Hurricane Air Quality Global Ensembles Meso Ensembles Real Time Ocean S/I Climate NCEP XXXXXXXX UKMO XXXXX ECMWF XXX
4
GFS CFS GFDL Hurricane SREF Eta Noah Land Surface Model Dispersion Air Quality 2005 NCEP Production Suite Atmospheric Model Dependencies Forecas t RUC GDASGDAS EDASEDAS WRF-NMM WRF-ARW ETA RSM L D A S GENSGENS Sev Wx WRF-NMM WRF-ARW
5
5 RUC GFS AnlHur GFS FcstNAM Fcst NAM Anl Waves SREF GENS
6
6 Performance Comparison Global NWP Gap widening for SH “Constant” gap for NH
7
7 Performance Comparison Hurricanes
8
8 wrt OIv2 1971-2000 climatology European Performance Comparison Seasonal Forecasts NCEP CFS CA (Statistical)
9
9 Performance Comparison BIASBIAS THREATTHREAT --- ECMWF --- UKM ___NCEP Global Models North American run THREATTHREAT 1993 1996 1999 2001 2003 2005 24 4872 Precipitation (24-72 h) 7/1/04-6/30/05
10
10 Improved Strategy Engage more partners from the US weather and climate community to: –Promote use of operational forecast systems by “non- operational” users Adopt community model concept to: –Supply improved diversity of scientific solutions –Enhance links and partnerships between research and operational communities –Support “Test Beds” which provide Technical support for codes and data More efficient transition to operations path based on results –Influence resource decisions based on operational research needs
11
11 Improved Strategy (cont) Examples –NASA-NOAA-DOD Joint Center for Satellite Data Assimilation –Real time ocean modeling –Supported community code (Data Assimilation) –WRF
12
12 NASA-NOAA-DOD Joint Center for Satellite Data Assimilation (JCSDA) –Multi-agency partnership (NOAA, NASA, DOD) –Mission Accelerate and improve the quantitative use of research and operational satellite data in weather and climate prediction models –Current generation data –Prepare for next-generation (e.g. NPOESS, METOP, COSMIC) instruments –Supports applied research Partners University, Government and Commercial Labs
13
13 Examples of Instrument-Specific Development at the JCSDA GPS Occultation (COSMIC) AIRS MODIS winds Surface emissivity for microwave instruments Advanced SST physical retrievals for IR & MW instruments A1 current New
14
14 US GODAE: Global Ocean Prediction with HYCOM Goal: to develop and demonstrate real-time, operational, high resolution ocean prediction systems for the Global Oceans and Basins NCEP Partners with University of Miami/RSMAS NRL Stennis, NRL Monterey, FNMOC NOAA PMEL, AOML Los Alamos National Laboratory Others (international, commercial) Hybrid isopycnal-sigma-pressure ocean model (called Hybrid Coordinate Ocean Model – HYCOM) Funded FY 2003-2007 by NOPP Chesapeake Bay
15
15 HYCOM Deployment Schedule North Atlantic World Oceans North-East Pacific Hawaii FY 2006 2007 2008 Global atmosphere-ocean Coupling and Hurricane-Ocean Coupling Initiate interactions with NOS on bay and estuary model boundary conditions; Initiate wave-current interactions. Storm Surge Modeling Ecosystem Modeling
16
16 Improved Strategy (cont) Data Assimilation Code –Gridpoint Statistical Interpolation (GSI) analysis Applied to global and regional (WRF) analyses Community-based (currently with minimum support) –46 users –NASA/GMAO has adopted code for their research and operations Ingests full suite of conventional and remotely-sensed (satellite and radar) observations –Community Radiative Transfer Model (CRTM) Contains advanced treatment of background errors Basis of advanced data assimilation techniques –High time and space density data –Simplified 4D-Var capability –Ready for ensemble information input
17
17 WRF Mesoscale community model: in development since 1997 – supported through USWRP, NOAA, DOD, FAA, UCAR, NSF Includes support for Boulder Development Testbed Center (DTC) and operational implementation at NCEP and DOD Currently supports same real-time code run at DTC and NCEP CMICMI NCAR ARW NCEP NMM Explicit Cores (e.g., Hurricane, Dispersion, Aviation)
18
18 WRF Implementation Schedule HiResWindow (Both cores): Implemented operationally at NCEP on 6/28/05 (~5 km) WRF SREF members: Operational FY05 (4 th Qtr) North American WRF: Operational in FY06 WRF SREF: Fully operational in FY07 Hurricane WRF: Operational in FY07* Rapid Refresh WRF: Operational in FY07* WRF Chem: Beyond 2008* * As resources allow
19
GFS CFS GFDL Hurricane SREF Eta Noah Land Surface Model Dispersion Air Quality 2005 NCEP Production Suite Atmospheric Model Dependencies Forecas t RUC GDASGDAS EDASEDAS WRF-NMM WRF-ARW ETA RSM L D A S GENSGENS Sev Wx WRF-NMM WRF-ARW
20
GFS CFS Hurricane WRF SREF WRF Noah Land Surface Model Dispersion Chem WRF* Air Quality 2007 NCEP Production Suite Atmospheric Model Dependencies Forecas t Rapid Refresh WRF GGSIGGSI RGSIRGSI WRF-NMM WRF-ARW ETA? RSM? L D A S GENSGENS Sev Wx WRF-NMM WRF-ARW *FY08
21
21 Summary NCEP, DOD, research community making progress on community model development and application –JCSDA –HYCOM –WRF –Community codes Need to build off this community effort and increased partnerships –Improve ongoing development and implementation process Work toward a full Earth System Modeling Framework for global and regional applications Ensure that the entire end-to-end effort is properly resourced
22
22 Backup
23
23 Earth System Modeling Framework 1.ESMF provides tools for turning model codes into components with standard interfaces and standard drivers 2.ESMF provides data structures and common utilities that components use i.to organize codes ii.to improve performance portability iii.for common services such as data communications, regridding, time management and message logging
24
24 NCEP, ECMWF, UKM Comparison Operating strategy ECMWFUKMETNCEPImpact Single forecast system with limited applications Single forecast system with international and domestic applications Multiple forecast systems with international and largest set of domestic applications Dilution of EMC management and scientific resources R:O Computing ratio ~4:1 Resources well planned UnknownR:O computing ratio ~1.3:1 Recent increased and reorganized support for computing Potential for improvement Operations department takes major role in optimizing, reviewing code Role of Operations Department unknown Operations department (NCO) does not review or optimize code EMC science resources diluted by software engineering
25
25 NCEP, ECMWF, UKM Comparison Organizational Factors ECMWFUKMETNCEPImpact Simpler chain of command (ECMWF Director to Council) Single management chain for operations & development Large management chain above EMC (NWS HQ, NOAA…) Research less coordinated with NOAA Dilution of EMC management resources; competition within NOAA for resources Does not have direct forecast responsibilities Forecast divisions & Field Offices NCEP Service Centers, Forecast regions and local offices Internal NWS competition for resources
26
26 NCEP, ECMWF, UKM Comparison Scientific Development & Community Relationships ECMWFUKMETNCEPImpact Operations drives research Weak influence over research community Little directed research to benefit operations; difficult transition of research to ops Recruits best scientists in Europe and U. S. – recruits U. S. scientists regularly Strong University collaborations (e.g. Reading U.; consortium for research aircraft) NCEP has little success recruiting top level U. S. scientists; with exceptions, NCEP community relationships weak Easier to recruit lower level scientists who require more management; best scientists direct many projects
27
27 Outreach (cont) NASA-NOAA-DOD Joint Center for Satellite Data Assimilation –Global wx: GMAO –Radiative transfer: NOAA/NESDIS, AER, U. Wisc. –Ocean data assimilation: U Md –Land surface modeling & data assim: NASA/HSB, Princeton U, U Wash.
28
28 Outreach (cont) –WRF Mesoscale wx (NAM, 2006): NOAA/FSL, NCAR, UOK, etc –WRF-ARW & WRF-NMM are “5 km” models in NCEP operations –WRF-ARW members in Short-Range Ensemble Forecast (SREF) Fall ’05 Hurricane (2007): URI, NOAA/AOML Rapid refresh (2007): NOAA/FSL Code convergence for mesoscale forecast systems SREF model diversity can be managed with minimum cost –Climate Test Bed S/I: GFDL, NASA/GMAO Unified forecast system –GFS: NOAA/FSL, U. Wisc. – hybrid coordinate model –Gridpoint Spectral Interpolation (GSI) analysis
29
29 Resource Comparison Ingredients for Improved Numerical Forecast Systems –Balance between Observations Data Assimilation & Model technology Computing resources –Processor growth equal to Moore’s Law –On-line disk proportional to processor capability –Archive proportional to processor capability –System support proportional to »Number of computers »Number of users Computing resources applied to –Operations –Integration and Testing upgrades
30
30 Computing Comparisons International Operational Weather & Climate Forecast Centers 2005-2006 WMO Survey (Majewski, 2005) Center (vendor, architecture) MP=Massively Parallel Peak Power (TF) Throughput (TF) NCEP (IBM, MP)161.0 UKMET (NEC, vector)51.5 ECMWF (IBM, MP)36.52.2 China (IBM, MP)211.3 Korea (Cray, vector)185.4 Japan (Hitachi, vector)288.4
31
31 Comparison of ECMWF and NCEP Operational Computing October 2004 ECMWFNCEP Processors2 x 21762 x 1280 Speed7.6 Gf6.8 Gf Sustained (6%)2 x 1 Tf2 x 0.5 Tf BudgetCpu - $10 M Tape,silo ~ $3.5 $11 M for all
32
32 EMC Mission In response to operational requirements: Maintain the scientific correctness and integrity of operational forecast systems –Adapt to format changes and other changing operational requirements –Adapt to new computing hardware –Monitor and ensure the integrity of operational observing systems Enhance (Test & Improve) Numerical Forecasts Through Advanced –Data assimilation techniques –Model physics (parameterizations) –Numerical methods –Computational efficiency Transition and Develop Operational Numerical Forecast Systems for: –Weather prediction (domestic, global, 1-15 days) –Ocean prediction (daily to annual, coastal to global) –Climate prediction (seasonal to inter-annual) Transition and Develop: transform & integrate code, algorithms, techniques from research status to operational status on NCEP computers Enhance: Test and improve NCEP’s numerical forecast systems via scientific upgrades, tuning, additional observations, in response to user requirements Maintain: Modify current operational system to adapt to ever-present external changes
33
33 FY05 EMC Budget Kelly report (2000) recommendation: 75% ORF, 25% “soft” 2002 budget supplement and adjustment: $2.8 M Total: $17.5 M <20022 201- 500 13 501- 800 3 >8013 Total41 Funding Sources
34
34 Human Resource Comparison UKMET twice NCEP for Global & Mesoscale development ECMWF 80% more than NCEP for Global development ECMWF Ops same as NCEP ops ECMWF covers computational efficiency and porting to new architecture Global & Mesoscale DevGlobal Dev Only Operations UKMETECMWF NCEP
35
35 Science plan for catching up Goal: to produce the highest forecast scores by 2010 –Synoptic scale forecasts –QPF –Hurricanes –Aviation –Marine & land transportation –Week 2 to S/I climate Advanced data assimilation methods –Better use of time dimension –Improved background covariances Flow dependence Ensemble methods –New development (with JCSDA) Clouds & precipitation Snow, Ice & polar regions Land & ocean surface –Adjoint of analysis system for improved tuning & understanding (with NASA)
36
36 Improved diagnostic analysis –Increased case study analysis –Work on “bust” cases –Greater effort on total system tuning Improved scientific development –Eddy simulation models* to advance PBL, stratus, stratocumulus parameterizations –Cloud resolving models* to advance cumulus and cloud fraction parameterizations –ECMWF, UKMET, NOGAPS initial states with GFS Higher resolution Ocean-Atmosphere-Land-Ice coupling Science plan for catching up (cont) * Currently done by UKMET
37
37 Science plan for catching up (cont) Enhanced ensemble systems –Data assimilation –Postprocessed, downscaled products –International ensemble system Enhanced community collaborations & outreach –Outreach Education on best use of products + Regular Workshops + –Full involvement in International Model intercomparison projects Field experiments + –Vigorous Visiting Scientist Program –WRF, global (weather & S/I climate) systems –USWRP, NSF research support + Currently done by ECMWF
38
38 Science plan for catching up (cont) Single forecast system –Applicable to global & mesoscale (nonhydrostatic) Software engineering group –Design systems which are easier to maintain –Improved software efficiency (so scientists don’t have to do it)
39
39 Organizational & Political Factors Organizational Factors ECMWFUKMETNCEPImpact Simpler chain of command (ECMWF Director to Council) Single management chain for operations & development Large management chain above EMC (NWS HQ, NOAA…) Dilution of EMC management resources; competition within NOAA for resources Does not have direct forecast responsibilities Forecast divisions & Field Offices Forecast divisions and WFOsInternal NWS competition for resources Management makes commitment to scientific plan, obtains funds NWP needs are lost in the budget process (NWS, NOAA, DOC, OMB, Congress). Initiatives have low success rate. No internal reallocation within NWS. Reliance on soft money to expand and maintain capabilities Operations department takes major role in optimizing, reviewing code Role of Operations Department unknown Operations department (NCO) does not review or optimize code EMC science resources diluted by software engineering Experienced staff culled from best of European Weather Services; Europeans better prepared in math, physics Entry level positions require years of training; Europeans better prepared in math, physics Entry personnel unfamiliar with operational NWP Center; “Black box” modelers require 1-3 years training Greater spinup time for new employees 57% “permanent”; 43% contractors Higher, tax free salaries 100% Civil Servants 38% Civil Servants; 62% contractors Increased management responsibilities on Civil Servants
40
40 Organizational & Political Factors (cont) Scientific Development & Community Relationships ECMWFUKMETNCEPImpact Operations drives research Weak influence over research community Little directed research to benefit operations; difficult transition of research to ops International reputation & prestige Some excellent scientific leaders Scientists well recognized & respected internationally but not nationally, in NOAA or NWS Little respect & recognition of EMC’s mission & capabilities; continually fighting critics Recruits best scientists in Europe and U. S. – recruits U. S. scientists regularly Strong University collaborations (e.g. Reading U.; consortium for research aircraft) NCEP has little success recruiting top level U. S. scientists; with exceptions, NCEP – OAR relationships weak Easier to recruit lower level scientists who require more management; best scientists direct many projects
41
41 ECMWF – NCEP(EMC,NCO, OD) comparison for global wx & climate ECMWFNCEP Total Employees206151 Ops employees (NCO)96101 Scientific staff7441 Director & infrastructure369 Total budget~$54 M$52.7* M * Clearly overestimated: NCEP OD, NCO + EMC (Global Wx & Climate)
42
42 Human Resource Comparison (cont) UKMET three times more than NCEP for Satellite Data Assimilation
43
43 EMC NCO R&D Operations Delivery Criteria Transition from Research to Operations Requirements EMC NCEP’s Role in the Model Transition Process OPS Life cycle Support Service Centers NOAA Research Concept of Operations Service Centers Test Beds JCSDA CTB WRF/Model JHT User Observation System Launch List – Model Implementation Process Field Offices Effort EMC and NCO have critical roles in the transition from NOAA R&D to operations Other Agency & International
44
44 EMC R&D Operations Delivery Criteria Requirements OPS Life cycle Support Service Centers Concept of Operations User Observation System 6. EMC Pre- Implementation Testing (Packaging and Calibration) 7. NCO Pre- Implementation Testing 8. Implementation Delivery 5. Level II:- Preliminary Testing (DA/Higher Resolution) 4. Level I:- Preliminary Testing (Lower Resolution) 3. Interface with Operational Codes 2. Code/Algorithm Assessment and/or Development 1. Identified for Selection 123 4 5 6 78 Launch List – Model Implementation Process NCEP’s (Modeling) Transition to Operations: Focus on EMC and NCO NCO EMC Effort Test Beds
45
45 Code/Algorithm Assessment and/or Development Transition Steps (Modeling) Identification for Selection 1 2 Interface with Operational Codes 3 Level I: Preliminary Testing (Lower Resolution) 4 Level II: Preliminary Testing (DA/Higher Resolution) 5 EMC Pre-Implementation Testing (Packaging/Calibration) 6 NCO Pre-Implementation Testing 7 Implementation/Delivery 8
46
46 Anomaly correlation for 5-day forecasts of 500hPa geopotential height Northern Hemisphere (20N-80N) Southern Hemisphere (20S-80S) O P E U Improvement From latest GFS Implementation (P)
47
47 NOAA Planning & Resource Allocation Example of new resources for expanded mission –Air Quality Contractor personnel Computing supplement NWS HQ management EMC management resources lacking NOAA Planning, Programming, Budgeting and Execution System (PPBES) –Environmental Modeling Program (EMP) –Addresses resources vs. forecast requirements Human Computing Observational
48
48 Additional Research Computing Former NCEP operational computer –1/3 current operational capability –Allocated to 3 major projects Satellite data assimilation Climate forecast development Advanced modeling (WRF, global) New NOAA “research” computer in procurement –October 2006 delivery –Replaces former NCEP operational computer
49
49 NCEP Global Model Strategy & ESMF Concept of operations –Single system for global and regional models –Performance permitting Migration to single model or Multiple dynamics and physics options in single structure –Single verification, observations data base obeying WMO standards –Single analysis code –All components ESMF compatible Enables multiple models and standard for coupling models Decreases code maintenance and code reuse Overall positive experience at Met Office & ECMWF
50
50 The Environmental Forecast Process Observations Analysis Model Forecast Post-processed Model Data Forecaster User (public, industry…) Numerical Forecast System Data Assimilation
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.