1. Examining Fresh Water Flux over Global Oceans in the NCEP GDAS, CDAS, CDAS2, GFS, and CFS P. Xie 1), M. Chen 1), J.E. Janowiak 1), W. Wang 1), C. Huang 1), C.-L. Shie 2), and L. Chiu 3) 1) NOAA Climate Prediction Center 2) UMBC / NASA Goddard Space Flight Center 3) GMU / NASA Goddard Space Flight Center

2. Objective:  To examine the oceanic precipitation and evaporation fields generated by NCEP reanalyses and climate model simulations through comparison with corresponding observations  Uncertainties in oceanic observations considered by using multiple data sources and taking into account the standard deviation among the different observations

3. Precipitation Observation Data  Three sets of observation data sets used:  CMAP / GPCP / TRMM  Annual climatology (mm/day) for 1988 – 2000  Largest uncertainties (standard deviation among observations) over ITCZ and high latitudes  Standard Deviation about 10% of the mean values

4. Evaporation Observation Data  Four products:  GSSTF2 / HOAPS3 J-OFURO-2 / SOC1.1a  Annual climatology (mm/day) for 1988 – 2000  Large uncertainties over tropical oceans, especially over E. Pacific and over oceans around the Maritime Continent  Standard Deviation about 10% of the mean values

5. E-P Annual Climatology  Based on 12 sets of E/P combinations  Standard Deviation about 10% - 20% of the mean values  Less uncertainties in zonal mean

6. DJF Precipitation Climatology  Spatial patterns reproduced well by all products  Pattern in GDAS in close agreement with observations  Orientation of SPCZ in CDAS1 too flat  Precipitation too heavy in GDAS, CDAS2, GFS and CFS  Atlantic ITCZ shirted southward and double ITCZ over E. Pacific in CFS

7. Precipitation Annual Cycle over Atlantic  30 o W-20 o W  Excessive preci- pitation in GDAS, CDAS2, GFS and CFS  Migration of ITCZ extremely well simulated in GDAS  ITCZ location shifted in CDAS2, GFS, and CFS  Southward shift of ITCZ reaches ~10 o during Spring in CFS

8. DJF Evaporation Climatology  Large-scale pattern reproduced reasonably well in all products  Over S.H., magnitude in all products close to that in the observations  Over N.H., excessive amount of evaporation compared to the observations

9. DJF E-P Climatology  Large-scale pattern reproduced reasonably well in all products  Larger magnitude in GDAS and CDAS2 over tropics  Excessive incoming E-P flux along Pacific ITCZ in GFS and CFS

10. Latitudinal Profile of Annual Mean  Location of maxima and minima well caught  CDAS1 reproduced the magnitude of E and P very well  All other products generated excessive amount of E / P over most latitudes

11. Anomaly Correlation for Precip.  Correlation very high (>0.8) between the two observations from 50 o S- 50 o N  Monthly anomaly reproduced very well by GDAS over most oceanic regions  Reasonable agreements for CDAS1 and CDAS2, especially over Pacific and North Atlantic

12. Standard Deviation of Precipitation Anomaly  Insufficient / excessive magnitude of inter- annual variations in precipitation simulated in CDAS1 / CDAS2  Results for GDAS not included due to short data record without major ENSO events

13. Anomaly Correlation for Evap.  Correlation very high (>0.8) between the two observations over extra- tropics and reasonably high (>0.6) over tropics  Evaporation anomaly reproduced well over extra- tropics but less desirable over tropics (r<0.5)  GDAS presents better agreements with the observations

14. Standard Deviation of Evaporation Anomaly  Distribution of anomaly standard deviation for CDAS1 compares well with the observations in both pattern and magnitude  Patterns generated by CDAS2 is similar to but magnitude is much larger than those in the observations

15. GDAS Performance  Time series of pattern correlation between GDAS simulation and observations (CMAP/HOAPS3)  Slight improvements in precipitation simulation from mid of 2003  Steady improvements in evaporation, especially over tropics

16. Summary  Seasonal variations and interannual variability of oceanic precipitation and evaporation in GDAS, CDAS1, CDAS2, GFS and CFS are broadly consistent with observations;  GDAS presents the best performance among the five products examined  However, certain consistent errors are seen:  The ITCZ is too weak in CDAS1 and too strong in the GDAS and GFS;  Shift in ITCZ / SPCZ positions, especially over Atlantic;  Systematic differences between GFS and CFS;  Excessive evaporation in GDAS, CDAS2, GFS and CFS

17. MAM Precipitation Climatology  Double ITCZ over E. Pacific reproduced well in GDAS, CDAS1, CDAS2 and GFS  ITCZ over E. Pacific (northern branch) weak and ITCZ over Atlantic shifted southward  ITCZ too strong over Pacific and Atlantic in GDAS, CDAS2, GFS, and CFS

18. JJA Precipitation Climatology  Spatial patterns reproduced well by all products  Pattern in GDAS in close agreement with observations  Precipitation too heavy in GDAS, CDAS2, GFS and CFS

19. JJA Evaporation Climatology  Large-scale pattern reproduced reasonably well in all products  Larger amount of evaporation in all products over Indian Ocean  Reasonable magnitude of evaporation elsewhere

20. JJA E-P Climatology  Large-scale pattern reproduced reasonably well in all products  Larger magnitude in GDAS and CDAS2 over tropics  Excessive/deficit incoming E-P flux over East / West Pacific ITCZ in GFS and CFS

21. DJF Evap. Anomaly Pattern [ENSO]  Regressional coeffi- cients of evaporation against NINO3.4  Magnitude of evapo- ration anomaly is much smaller (~1/5) than that of precipitation  Anomaly pattern well simulated by CDAS1, CDAS2 and GFS, especially over Atlantic  Anomaly pattern poorly simulated by CFS over Pacific