Design of the 30-year NCEP CFSRR
T382L64 Global Reanalysis and T126L64 Seasonal Reforecast Project (1979-2009)
Suru Saha and Hua-Lu Pan, EMC/NCEP
With Input from Stephen Lord, Mark Iredell, Shrinivas Moorthi, David Behringer, Ken Mitchell, Bob Kistler, Jack Woollen, Huug van den Dool, Catherine Thiaw and others
Operational Daily CFS
• 2 9-month coupled forecasts at 0000GMT • (1) GR2 (d-3) atmospheric initial condition • (2) Avg [GR2(d-3),GR2(d-4)] • T62L28S2 – MOM3 • GODAS (d-7) oceanic initial condition 2004 CFS Reforecasts The CFS includes a comprehensive set of retrospective runs that are used to calibrate and evaluate the skill of its forecasts. Each run is a full 9-month integration. The retrospective period covers all 12 calendar months in the 24 years from 1981 to 2004. Runs are initiated from 15 initial conditions that span each month, amounting to a total of 4320 runs. Since each run is a 9-month integration, the CFS was run for an equivalent of 3240 yr! An upgrade to the coupled atmosphere-ocean-seaice-land NCEP Climate Forecast System (CFS) is being planned for Jan 2010.
Involves improvements to all data assimilation components of the CFS:
• the atmosphere with the new NCEP Gridded Statistical Interpolation Scheme (GSI) and major improvements to the physics and dynamics of operational NCEP Global Forecast System (GFS)
• the ocean and ice with the NCEP Global Ocean Data Assimilation System, (GODAS) and a new GFDL MOM4 Ocean Model
• the land with the NCEP Global Land Data Assimilation System, (GLDAS) and a new NCEP Noah Land model For a new Climate Forecast System (CFS) implementation Two essential components: A new Reanalysis of the atmosphere, ocean, sea-ice and land over the 31-year period (1979-2009) is required to provide consistent initial conditions for: A complete Reforecast of the new CFS over the 28-year period (1982-2009), in order to provide stable calibration and skill estimates of the new system, for operational seasonal prediction at NCEP There are 4 differences with the earlier two NCEP Global Reanalysis efforts:
1. Much higher horizontal and vertical resolution (T382L64) of the atmosphere (earlier efforts were made with T62L28 resolution)
2. The guess forecast will be generated from a coupled atmosphere – ocean – seaice - land system
3. Radiance measurements from the historical satellites will be assimilated in this Reanalysis
4. 4 Soil layers – Noah Land Data Assimilation system Progress GR1/GR2 2007 GDAS (CFSRR)
Vertical structure sigma sigma-pressure Spectral resolution T62 T382 Horizontal resolution 210 km 35 km Vertical layers 28 levels 64 levels top 3 hPa .266 hPa Layers above 100 hPa 7 24 Layers below 850 hPa 6 13 Lowest layer 40 m 20 m Soil Layers 2 4 3D-var SSI GSI NESDIS data retrievals radiances Vertical coordinate comparison across North America
9New format and improved SBUV ozone data 9Calibrated 3 ch. SSU radiance data (corrected for cell pressure leaks) CFSRR at NCEP
Climate Forecast System
GDAS 6hr Atmospheric Model GSI GFS (2007) T382 L64 S4 24hr 6hr LDAS Land Model Ice Model Ice Anl Ocean Model MOMv4 6hr GODAS fully global 3DVAR 1/2ox1/2o (1/4o in tropics) 40 levels CFSRR BOARD MEETING PRESENTATIONS
NCEP 2007 November 7-8 http://cfs.ncep.noaa.gov/cfsreanl/PRESENTATION/20071107_board_meeting.html
• Overview of the CFSRR Suru Saha, EMC NCEP • The Atmospheric Model Shrinivas Moorthi, EMC NCEP • The Atmospheric Data Assimilation Russ Treadon, EMC NCEP • GPS Radio-Occultation Data Lidia Cucurull, EMC NCEP • The Ocean Model Jiande Wang, EMC NCEP • The Ocean Data Assimilation Dave Behringer, EMC NCEP • The Sea Ice Model Xingren Wu, EMC NCEP • The coupler Jun Wang, EMC NCEP • The Sea Ice Analysis Bob Grumbine, EMC NCEP • The Sea Surface Temperature Analysis Diane Stokes, EMC NCEP • The Snow Analysis George Gayno, EMC NCEP • The Land Surface Model and Land Surface Data Assimilation Ken Mitchell, EMC NCEP • The Data Jack Woollen, EMC NCEP • Tropical Cyclones Bob Kistler, EMC NCEP • Ozone Craig Long, CPC NCEP • SSU Mark Liu, NESDIS • MSU Cheng-Zhi Zou , NESDIS • Satellite Bias Correction Haixia Liu , EMC NCEP • The Results Hua-Lu Pan , EMC NCEP • EMC Monitoring Cathy Thiaw , EMC NCEP • A review of the CPC's Monitoring Efforts for the CFS Reanalysis Muthu Chelliah , EMC NCEP The changing number and characteristics of observations ONE DAY OF REANALYSIS 12Z GSI 18Z GSI 0Z GSI 6Z GSI
0Z GLDAS
12Z GODAS 18Z GODAS 0Z GODAS 6Z GODAS 9-hr coupled T382L64 forecast guess (GFS + MOM4 + Noah)
1 Jan 0Z 2 Jan 0 Z 3 Jan 0Z 4 Jan 0Z 5 Jan 0Z
2-day T126L64 coupled forecast ( GFS + MOM4 + Noah ) 4 Simultaneous Streams
• Jan 1979 – Oct 1989 11 years • Apr 1989 – Oct 1998 10 years • Apr 1998 – Oct 2004 7 years • Apr 2004 – Dec 2009 6 years
6 month overlap for ocean and land spin ups
Reanalysis to cover 31 years (1979-2009) + 21 overlap months
Reforecasts to cover 28 years (Jan 1982 – Dec 2009) SOME NOTES PROPOSED TIME LINE FOR COMPLETION OF CFSRR •January to December 2008: Begin Production and Evaluation of the CFS Reanalysis for the full period from 1979 to 2008 (30 years) •January to December 2008: Begin running CFS Retrospective Forecasts for 2 initial months: October and April, and evaluate the monthly forecasts as well as the seasonal winter (Lead-1 DJF) and summer (Lead-1 JJA) forecasts. •January to October 2009: Continue running the CFS Reforecasts (for the rest of the 10 calendar months) •November 2009: Begin computing calibration statistics for CFS daily, monthly and seasonal forecasts.
January 2010: Operational implementation of the next CFS monthly and seasonal forecast suite. NOAA Collaborators
• NOAA/CPC is actively involved in the monitoring of the Reanalysis • NOAA/NCDC is actively involved in the archival of the Reanalysis. DATA ARCHIVAL • NCDC/NOAA will archive and distribute both the CFS Reanalysis and Reforecasts, through their NOMADS system. • Targets: • CFS Reanalysis - mid 2009 • Reforecast data - mid 2010. Data to be archived and available from NCDC
CFS Hi Res IC (6-hourly) T382L64 Atmosphere + .5 degree Ocean
CFS Lo Res IC (6-hourly) T126L64 Atmosphere + 1 degree Ocean
Full Ingest Data (6-hourly) All input data for the re-analysis pgbh 0.5 x 05 (Hourly) 37 standard pressure level atmosphere products flxf T382 Gaussian (Hourly) surface and radiative fluxes on model grid ocnh 0.5 x 0.5 (Hourly) 40 standard depth level ocean products diabf 1.0 x 1.0 (Hourly) 37 standard pressure level diabatic heating rates ipvh 0.5 x 0.5 (Hourly) 16 standard isentropic level atmosphere products Status
• New format and improved SBUV ozone data • Calibrated SSU radiance data (corrected for cell pressure leaks) has been tested for impact • High resolution SST analysis has been tested for impact • Satellite radiance bias correction estimated for each new satellite with a 3-month run of the full coupled GDAS at T382L64. Status (contd)
• A mini T62L64 atmospheric-only Reanalysis is completed to pass through all atmospheric data from 1979 – 2007 • We had to adjust the streams as the later year data amount is much larger than earlier ones • Results of 5-day forecasts made from every 0Z cycle are encouraging. Thank you….
Email : [email protected]
Website : http://cfs.ncep.noaa.gov Mingyue Chen and UnifiedUnified DailyDaily GaugeGauge DataData Pingping Xie Dense gauge networks from special CPC collections over US, Mexico, and S. America; GTS gauge network elsewhere Daily reports available from ~17,000 stations Precip blending 28-year CFS re-forecast archive by NCDC (1982-2009) 6-hourly Pgb and Flx * 37 standard pressure level atmosphere products 6-hourly Ocn * 40 standard depth level ocean products 6-hourly Ipv * 16 standard isentropic level atmosphere products
• * 1.0 x 1.0 for first 6 months of forecast ; 2.5 x 2.5 for next 6 months of forecast 37 Pressure (hPa) Levels : pgb, ipv, egy and diab (atmosphere) 1000 975 950 925 900 875 850 825 800 775 750 700 650 600 550 500 450 400 350 300 250 225 200 175 150 125 100 70 50 30 20 10 7 5 3 2 1 40 Levels (depth in meters) in ocn (ocean)
4478 3972 3483 3016 2579 2174 1807 1479 1193 949 747 584 459 366 303 262 238 225 215 205 195 185 175 165 155 145 135 125 115105 95 85 75 65 55 45 35 25 15 5
16 Isentropic (K) Levels
270 280 290 300 310 320 330 350 400 450 550 650 850 1000 1250 1500 A Brief History of Conventional Data Preparation for Reanalysis at NCEP
•TEMP PILOT AIREP PIREP ACARS SATOB SYNOP MARINE BOGUS
•R1/R2 merged data sources from NCAR, NCDC, NCEP, JMA, and ECMWF
•R1/R2 merged qc’d datasets were packaged and delivered to the JMA-25 project
•R1/R2 datasets were reprocessed and enhanced by NCEP/NCAR for ERA-40
•NCEP obtained the ECMWF MARS archive of conventional data, 1979-1994
•NCAR improved versions of key datasets, ie, COADS, ON29/RAOBS, ON124
•R1/R2 and new datasets were merged for the North American Regional Reanalysis
•The latest versions of all source datasets were merged for the MERRA project
•Much effort was made by MERRA people to find data gaps and other problems
•CFSRR will use conventional observations prepared for MERRA, 1978-2000
•CFSRR will use NCEP/NCO operational archives to overlay and complete 1997-2007 SNOW
• Current GFS OPS updates snow pack once per day (00z). – SNO2MDL program maps NESDIS IMS and AFWA snow analysis data to physics grid. – Output from SNO2MDL ingested by SFC- CYCLE program. • Snow updated on model grid via direct replacement. • CFSRR will follow similar procedure. NESDIS IMS Data
• IMS – Interactive Multisensor Snow and Ice Mapping System. – NESDIS-SSD analyst uses satellite imagery and snow detection algorithms to determine snow cover over NH once per day. – Cover is yes/no flag. – 4 km resolution since Feb. 2004. – Product started Feb. 1997 (23 km resolution). – NIC in Suitland assumes responsibility for IMS starting Feb. 2008. AFWA Snow Depth Data • AFWA SNODEP model produces global snow depth analysis once per day (47 km resolution). – Uses microwave-based detection algorithms, in-situ reports of snow depth, and its own climatology. – Analyst adjusts analysis based on satellite imagery in NH each day and in SH twice per week. • Product has long history: all years of CFSRR. – 1979-1998 acquired from NCDC by Mike Fennessy (COLA), who Gribbed it with help from EMC. • EMC filled some gaps and removed several “bad” analyses. – 2002 to present in Air Force non-grib format used in present ops (via EMC archive of Jesse Meng) – 1999-2001 extracted from NCO run history files. radiance NOT used radiance used
……
(Reference) Tropical Cyclone (TC) Review •Tcvitals – Worldwide observations of TC • location, size, strength, and movement – Merged dataset • NCEP archive 1997 -> • Fiorino’s best tracks 1978 -2006 • Relocation – predominant – TC in guess is located and bodily moved to tcvital location • Bogus vortex winds – If TC not found in guess, bogus vortex wind profiles are created to “spin-up” storm in GSI • <1997 Fiorino’s bogus winds • >2001 Steve Lord’s synthetic bogus winds • 1997-2001 – mix depending on Tcvitals storms details 500 mb 5 Day Forecasts
90
80 NH GFS SH GFS NH CDAS/Reanl SH CDAS/Reanl NH CFSRR T62 70 SH CFSRR T62
60
50
40 1982 1987 1992 1997 2002 2007 Climate Forecast System Reanalysis and Reforecast (CFSRR) • Reanalyis – 1979-2008 – 2008 GDAS with GSI – Coupled Ocean Data Assimilation (+ice) – Coupled Land Data Assimilation with daily global precipitation analysis • Reforecasts – Calibrate upgrade to NCEP CFS (Climate Forecast System) Jack Woolen and others have spent years improving the data base of conventional observations --much more complete than before --errors better understood
Great deal of experience now with satellite bias corrections
Experienced with changes in observations in last 10 years Knowledge is being applied to new reanalysis
GFS produces much more skilled forecasts than CDAS --GFS has proven track record in forecasting hurricane tracks and in seasonal forecasts as CFS, indicating that GFS produces much more realistic tropical atmosphere than CDAS in both analyses and forecasts CFSRR
Execution: Jan-Dec 2008 Reanalysis :4 concurrent streams Reforecasts : April, Oct
2008 1979
IBM RISC
2009 2010 Execution: Jan-Oct 2009 Reforecasts : other 10 months Production - 2010 Purpose of the CFSRR 1. The reanalysis is intended to provide the best initial conditions for the CFS reforecast 2. The reanalysis can also be used to address calibration and statistical applications for weather and monthly forecasts 3. The reanalysis and reforecast can be used to address the coupled vs uncoupled forecasts from day one onward CFSRR at NCEP
Climate Forecast System
GDAS 6hr Atmospheric Model GSI GFS (2007) T382 64 levels 24hr 6hr LDAS Land Model Ice Model Ice Anl Ocean Model MOMv4 6hr GODAS fully global 3DVAR 1/2ox1/2o (1/4o in tropics) 40 levels For a new CFS implementation (contd)
1. Analysis Systems : Operational GDAS: Atmospheric (GDAS)-GSI Ocean-ice (GODAS) and Land (GLDAS) 2. Atmospheric Model : Operational GFS 3. Ocean Model : New MOM4 Ocean 4. Land Model : Operational Noah Land Model 5. Sea Ice Model: New Sea Ice Model For a new CFS implementation (contd) 1. An atmosphere at high horizontal resolution (spectral T382, ~38 km) and high vertical resolution (64 sigma- pressure hybrid levels) 2. An interactive ocean with 40 levels in the vertical, to a depth of 4737 m, and high horizontal resolution of 0.25 degree at the tropics, tapering to a global resolution of 0.5 degree northwards and southwards of 10N and 10S respectively 3. An interactive sea-ice model 4. An interactive land model with 4 soil levels CFSRR T62L64 Atmospheric Only Reanalysis 5 Day fcst scores vs CFSRR fcsts from GR2
500 mb 5 Day Forecasts
90
80 NH CFSRR T62 SH CFSRR T62 Annual Avg NH CFS R2 CFS R2
70 1982-2005 CFSRR R2 60 NH 74.6 72.3 50 SH 71.1 63.3
40 1982 1987 1992 1997 2002