raotalk3

Slide1: 

Global Air-Sea Flux Data Sets Glenn White Global Climate and Weather Modeling Branch EMC/NCEP NWS/NOAA/DOC Glenn.White@noaa.gov

Slide2: 

Sources of air-sea fluxes: --COADS ship observations --satellite-based estimates SRB surface radiation estimates --numerical weather prediction data assimilation operational reanalyses --FSU surface winds, stress --Yu, Sun and Weller-objective analysis of ships, buoys, satellite retrievals, NWP output

Slide3: 

1) WCRP/SCOR Working Group on Air-Sea Fluxes— a) Report on our current knowledge of air-sea fluxes b) Workshop proceedings http://www.soc.soton.ac.uk/JRD/MET/WGASF Or from WMO/WCRP New group being organized 2) SEAFLUX-emphasis on air-sea fluxes from satellites http://paos.colorado.edu/~curryja/ocean 3) SURFA—Peter Gleckler and Jan Polcher Comparison of operational NWP centers’ surface fluxes (land and ocean) to high quality in-situ measurements

Slide4: 

Fluxes from COADS data Da Silva et al.--UWM Josey et al.--SOC Careful corrections applied to ship data Global imbalance Not truly global coverage  

Slide5: 

(left) Correlation of monthly mean evaporation from NCEP1 reanalysis with COADS;(right) number of COADS ship reports/month Results similar for sensible heat, net heat flux and wind stress Reanalyses agree well with COADS where COADS has data. Reanalyses have observations other than ship data. Indicates that reanalyses’ latent and sensible heat and wind stress are reasonable.

Surface radiation budget (SRB-1) Jul. 83-June 91Satellite measurements TOAISCCP cloudsRadiative transfer modelSatellite-based short wave estimates often higher than surface observations (BSRN)SRB NLW Global mean of 48 W/m2Kiehl and Trenberth (1997) 66 W/m2SRB Net radiation fairly large, implying larger evaporation than any other estimate: 

Surface radiation budget (SRB-1) Jul. 83-June 91 Satellite measurements TOA ISCCP clouds Radiative transfer model Satellite-based short wave estimates often higher than surface observations (BSRN) SRB NLW Global mean of 48 W/m2 Kiehl and Trenberth (1997) 66 W/m2 SRB Net radiation fairly large, implying larger evaporation than any other estimate

Slide7: 

Global Mean Ocean Heat Budget

Slide8: 

Best way to assimilate satellite data is to assimilate what satellite actually measured into system that considers all other data and applies laws of physics—an atmospheric data assimilation system Independent estimates of surface fluxes needed to keep data assimilation estimates honest

Slide9: 

Precipitation Estimates Global Precipitation Climatology Project CMAP (Xie and Arkin) Based on rain gauges, satellite-based estimates OPI—quick-look CMAP Magnitude over ocean uncertain Satellite algorithms may be more tuned to tropics than midlatitudes

Slide10: 

(CMAP) Which is correct?

Slide11: 

lower bound on uncertainty in precipitation CMAP Precip 1988-92 GPCP-CMAP

Slide12: 

NCEP-1 with CMAP NCEP-2 with CMAP Monthly mean precipitation anomalies from NCEP reanalyses correlated with CMAP

Slide13: 

For best air-sea fluxes, need reliable estimates of fluxes and near-surface fields at regular spatial and temporal intervals a few hours after observation time produced from vast amount of quality-controlled data carefully interpolated in space and time by laws of physics, in which information about surface fields is extrapolated from other fields, carefully monitored continuously by diverse, critical users, rigorously tested for its accuracy.   That’s data assimilation for NWP. Forecasting a system's behavior is a tough test (not the only test) of how well you understand that system.

Slide14: 

Problems: "spin-up"--initial adjustment of fields ERA-15 recommended use of 12-24h forecasts for fluxes, not 0-6.  NWP systems continually being improved. Discontinuities introduced, can be significant in fields more defined by model physics than by observations. Changes to improve forecasts, not to improve air-sea fluxes Timeliness requirement may limit use of best parameterizations, highest resolution, relevant data.

Slide15: 

Spin-down in tropics; more like OPI Spin-up in mid-latitudes; less like OPI Operational GFS

Slide16: 

In May 2001 in operational GFS, prognostic cloud liquid water replaced diagnostic cloudiness based on relative humidity. New aerosol climatology The changes increased forecast skill, reduced model biases and introduced more physical cloudiness. Surface net short wave is now too high; new cloudiness needs adjustment. New cloud microphysics is being tested. Global mean DSW increased from 198 to 211 W/m2.

Slide17: 

New-Old Operational GFS

Slide18: 

Increased resolution, minor analysis changes in late Oct. 2002 increased global mean precipitation from 3.02 to 3.22 mm/day.

Slide19: 

NCEP Uses analyzed SST, currently averaged over a few days No diurnal cycle in SST doesn’t use TOGA-COARE parameterization doesn’t adjust ship obs for ship height observations at island station assimilated as if typical of ocean Output every 6 hours No wave model in GFS

Slide20: 

Reanalysis--use frozen data assimilation system to reanalyze past and present meteorological data, producing long record uncontaminated by changes in system, but still contaminated by changes in observations.  

Slide21: 

Reanalyses available 1) NCEP-1 1995 model, T62, 1948-present CDAS 2) NCEP-2 1979-present Same resolution, analysis system, almost same data Corrected mistakes Updated physics  3) ERA-15 12-24 hr twice daily forecasts to avoid "spin-up" 1979-93, T106 4) ERA-40 higher resolution, improvements, esp. to land surface In progress  

Slide22: 

NCEP1--stratus clouds in eastern subtropical oceans stress too weak in equatorial Pacific ocean surface albedo wrong discontinuity at beginning of 1979 NCEP2--reasonable stress in equatorial Pacific used to produce reasonable ocean reanalysis at NCEP correct ocean surface albedo lack of stratus clouds in eastern oceans ERA15--lack of stratus clouds in eastern oceans discontinuities due to problems with satellite radiances best precip pattern of first reanalyses ERA40--problems with radiances due to Pinatubo

Slide23: 

Reanalyses have problems with surface radiative fluxes. —cloudiness and moisture poorly known and modeled. Reanalyses’ sensible and latent heat and wind stress are more realistic. --Parameterizations dependent on near-surface fields, which NWP data assimilation can infer from atmospheric fields. --Reanalysis estimates may be as realistic as any global fields of sensible and latent heat and wind stress CMAP precip more realistic than reanalyses, except possibly over high latitude oceans

Slide24: 

Surface fluxes—GDAS(current system) vs. CDAS (NCEP1) GDAS—higher resolution better evaporation at high wind speed better precipitation pattern better wind stress in equatorial Pacific Corrected ocean surface albedo Precipitation and evaporation greater Sensible heat less Surface energy balance more out of balance

Slide25: 

5S-5N Pacific FSU 1980-93 SOC Time-mean Us NCEP-1 too weak; NCEP-2 better COADS NCEP-1

Slide26: 

Anomaly correlation with FSU--Pacific NCEP-1 NCEP-2 Zonal surface stress

Slide27: 

NCEP-1 minus SRB-1 Surface NSW 1985-1991

Slide28: 

Net short wave over No. Pacific in July from SRB-1 and 3 reanalyses. ERA-15, NCEP-2 show no evidence of stratus clouds. SRB-1 ERA-15 NCEP-1 NCEP-2 Surface NSW July 1983-90

Slide29: 

Operational SRB-1 Surface NSW July 03 July 1983-90

Slide30: 

Stratus displaced from coast in operational Air Force nephanalyses Operational Low cloud cover July 2003

Slide31: 

Operational OPI (quick look CMAP) NCEP-1 Precipitation DJF 02/03

Slide32: 

NCEP-1 Operational COADS Evaporation DJF 02/03

Slide33: 

Operational NCEP-1 COADS Sensible heat flux DJF 02/03

Slide34: 

Operational NCEP-1 SRB-1 Surface downward short wave DJF 2002/2003

Slide35: 

NCEP-1 Operational SRB-1 Surface NSW DJF 02/03

Slide36: 

Net heat flux into ocean DJF 02/03 Operational CDAS COADS

Slide37: 

My preferences Precipitation--CMAP or GPCP Net Short Wave--SRB Sfc. Stress--NCEP2, ERA or FSU Sensible heat--NCEP1 or ERA Latent heat--NCEP1, NCEP2 or ERA Net heat flux--combine SRB NSW and reanalyses

Slide38: 

Need to improve knowledge of air-sea fluxes, especially clouds, magnitude of hydrological cycle, atmospheric moisture. Cloudiness major problem in NWP fluxes. Test air-sea fluxes by using different data sets in ocean models

Time for another NCEP global reanalysis?Is operational system ready for the last 56 years?How to reduce effects of changes in observations?: 

Time for another NCEP global reanalysis? Is operational system ready for the last 56 years? How to reduce effects of changes in observations?

Slide40: 

Global Mean Balances Aug02-Jul03  

Slide41: 

Zonal mean net short wave over ocean Reanalyses much different from satellite estimate.

Slide42: 

Zonal mean net heat flux into ocean DJF 02/03 COADS NCEP-1 Operational