Maria Valdivieso Department of Meteorology, University of Reading, UK ▶ Focus on surface heat fluxes ▶ Timeseries comparison at buoy sites
Products Surface Heat Flux Assimilation Heat Flux SSTTime Frame ECMWF-ORAS4 XX JMA-MOVEG2 XXX JMA-MOVECORE XXX Kiel-GECCO2 XX NCEP-GODAS XX Reading-UR025.4 XXX Met Office-GloSea5 XX Mercator-GLORYS2V1 XXX GFDL-ECDA (Coupled) XX NCEP-CFSR (Coupled) XX LEGI-MJM95 (Control) XX Monthly mean data regridded onto the WOA grid Common period 2004 – 2009 ▶ Other flux data sets: ▶ Other flux data sets: ISCCP+OAFlux, NOC2.0, ERAI, NCEP –R2
Global Integrals Surface heat fluxes from ocean reanalysis products averaged over the global ocean (common ocean-land mask). Fluxes are positive into the ocean. Units are in Wm -2
Products1993 – – 2009 ECMWF-ORAS44.99 ± ± 0.47 JMA-MOVEG27.90 ± ± 1.97 JMA-MOVECORE8.60 ± ± 0.50 Kiel-GECCO20.74 ± ± 0.37 NCEP-GODAS0.10 ± ± 1.29 Reading-UR ± ± 1.35 Met Office-GloSea58.95 ± ± 2.01 Mercator-GLORYS2V12.87 ± ± 1.62 GFDL-ECDA (Coupled)−14.60 ± 1.28 NCEP-CFSR (Coupled)13.96 ± ± 1.06 LEGI-MJM95 (Control)0.30 ± ± 0.56 ISCCP + OAFlux25.04 ± ± 0.67 NOC ± ± 2.63 ECMWF-ERAI7.41± ± 1.32 NCEP-R21.66 ± ± 1.77 Fluxes are positive into the ocean. Units are in Wm -2 ◀ ◀ ◀ ◀
Mean
Equatorial heating in the middle of the range span by other products Warming pattern in the SH underestimated at all latitudes Heat loss north of 20N seems reasonable
Also shown are the zonal integrals from Ganachaud and Wunsch (2003) and Lumpkin and Speer (2007) obtained from inverse analysis of WOCE sections MHT as inferred from estimated surface heat fluxes The mean transport due to the net heat uptake is ~ 2 PW, representing a non-zero heat storage in these ocean reanalyses Mean
ISCCPERAINOC2.0ECDA ± ± ± ± 0.41 Global averages over 2004 –2009. Units in Wm -2 Mean Cloud estimation from ships Satellite-based Atmos reanalysis Coupled reanalysis
HadISST (Rayner et al., 2003) combines in situ + satellite-based data Mean Using NCEP-R2 + relaxation to weekly OISST Using NCEP surface fields Using COREII surface fields JMA Reanalysis surface fields
Using corrected ERAI Qsw Assimilating Reynolds SSTs OSTIA + real time GHRSST after 2008 Coupled Reanalyses SST Diff Maps (cont.) Mean
Differences in the bulk fluxes Eddy-permitting + Core bulk forcing using ERAI surface fields Assimilating EN3 T/S profiles Assimilating SST + EN3 data Assimilating SST + CORA data No data assimilation Mean
CLIMODE: 38.5N, 65W KEO: 32.4N, 144.5E WHOTS: 22N, 158W NTAS: 15N, 51W TAO_w: 0, 165E TAO_e: 0, 110W STRATUS: 25S, 85W CLIMODE: 38.5N, 65W KEO: 32.4N, 144.5E WHOTS: 22N, 158W NTAS: 15N, 51W TAO_w: 0, 165E TAO_e: 0, 110W STRATUS: 25S, 85W The underlying map is the annual mean (1993 – 2009) net surface heat flux from the OAFlux + ISCCP product (Wm -2, positive downward) available at
CLICLMO DE ORAS4MOVECGECCOGODASUR025.4GloSea5GLORYSMJM95CFSRECDA ENSEMBLE emble MEAN Mean DIFF STD Monthly Climatology Mean (ISCCP+OAF) = ±22.4 Wm -2
Sea Surface Temperatures Interestingly, GODAS net heat flux is much too weak here (only -47 Wm -2 ), yet the SST is reasonably well reproduced. GECCO2 is systematically too cold; MOVECORE is too warm warm winters in 2008 and 2009 Monthly Climatology
Mean (ISCCP+OAF) = ±16.2 Wm -2 KEOSORAS4MOVECGECCOGODASUR025.4GloSea5GLORYSMJM95CFSRECDA ENSEMBLE MEAN Mean DIFF STD
Mean (ISCCP+OAF) = ± 9.96 Wm -2 WHOTSORAS4MOVECGECCOGODASUR025.4GloSea5GLORYSMJM95CFSRECDA Ensemble MEAN DIFF STD
Sea Surface Temperatures Surface Heat Fluxes
Mean (ISCCP+OAF) = ± 8.2 Wm -2 NTASORAS4MOVECGECCOGODASUR025.4GloSea5GLORYSMJM95CFSRECDA ENSEMBLE MEAN DIFF STD
Mean (ISCCP+OAF) = ±8.83 Wm -2 STRATUSORAS4MOVECGECCOGODASUR025.4GloSea5GLORYSMJM95CFSRECDA ENSEMBLE MEAN Mean DIFF STD
Sea Surface Temperatures Surface Heat Fluxes
Mean (ISCCP+OAF) = ±21.62 Wm -2 STRATUSORAS4MOVECGECCOGODASUR025.4GloSea5GLORYSMJM95CFSRECDA ENSEMBLE MEAN DIFF STD
Mean (ISCCP+OAF) = ± 9.86 Wm -2 STRATUSORAS4MOVECGECCOGODASUR025.4GloSea5GLORYSMJM95CFSRECDA Ensem ble MEAN DIFF STD
Mean ( ) differences at buoy sites ISCCP + OAFlux versus Ocean Reanalyses Ocean Reana Based minusISCCP+OAFlux CLIMODE For CLIMODE, the overall biases (less cooling) result primarily from the winter months. Here, fluxes are sensitive to the model resolution. WHOTS NTAS In the central tropical Pacific (WHOTS) and tropical Atlantic (NTAS), fluxes show less warming during spring and summer. STRATUS In the south east Pac (STRATUS), fluxes show less warming in the summer months and more cooling in the winter. TAO For TAO locations, fluxes provide less warming all year round. KEO TAO_w KEO, for the Kuroshio region, and TAO_w have the smallest differences.
Comparing with other flux products Mean Differences Mean Differences Product - (ISCCP + OAFlux) Ocean Reana Based minusISCCP+OAFlux
Most ocean reanalysis show a positive imbalance in global surface heating (ensemble mean of ~ 4 Wm-2 over ). This can be as large as 14 Wm-2 in coupled reanalyses. Generally, the imbalance is reduced as more observations become available after Ocean reanalysis-based fluxes are biased low compared to ISCCP+OAFlux data at all buoy locations. Variability is generally well reproduced. The result that the reanalysis SSTs compare reasonably well with HadISST data while the reanalysis-based fluxes are systematically too low compared to ISCCP+OAFlux data suggests that the models stratify the upper ocean too strongly. This may be a result of inadequate vertical mixing, weak advection,... Direct flux measurements are needed for further validation. Summary