Uncertainties in reanalysis surface wind stress and their relationship with observing systems
- 84 Downloads
Atmospheric reanalysis surface wind stress (SWS) estimates have been widely used as surface forcings to drive ocean model simulations and ocean reanalyses, and for understanding climate variability. In this study, we quantify uncertainty in SWS products from six reanalyses, with five from the third generation reanalysis (CFSR, MERRA-1, MERRA-2, JRA-55, ERA-Interim) and one from the first generation reanalysis (R2). Our goals are to (1) characterize uncertainties in monthly mean reanalysis SWS estimates, (2) investigate relationship between the spatial and temporal variations of uncertainties and changes in numbers of observed surface wind data and (3) examine consistency of SWS estimates across different generation of reanalysis products. The six reanalysis SWS estimates broadly agree on the global pattern but differ considerably in magnitude. Compared with TAO SWS estimates, R2 and MERRA-1 significantly underestimate SWS speed over the central Pacific Ocean, while MERRA-2 overestimates easterly wind stress over the north western Pacific. On interannual time scale, high consistency among the reanalyses is located in mid-to-high latitudes, while large uncertainty is found in tropical oceans. All six reanalyses exhibit a similar pattern of wind response to ENSO, but differ significantly in the strength. Compared with TAO estimates, MERRA-2, JRA-55 and ERA-Interim capture the evolution and amplitude of wind anomaly associated with ENSO reasonably well, while CFSR is the outlier which underestimates the peak wind anomaly by more than 35%. Our analysis highlights the importance of observed surface wind data in constraining the reanalysis SWS. It is found that the time variations of the spread among the six reanalyses follows closely to changes in the input of ocean surface winds from satellite scatteometers in all the three tropical oceans. In the tropical Pacific, the accuracy of reanalysis SWS is also linked to the return number of TAO/TRITON winds. The correlation between R2 and the third generation reanalysis products display a clear TAO signature, while the third generation reanalysis products are in a closer agreement with each other.
The authors thank Dr. Wesley Ebisuzaki for clarifying wind observations input in the R2 and CFSR reanalyses. TAO/TRITON wind stresses were downloaded from https://www.pmel.noaa.gov/tao/oceansites/flux/main.html. MERRA-1 and MERRA-2 data were from the NASA Goddard Earth Sciences (GES) Data and Information Services Center (DISC) at http://disc.sci.gsfc.nasa.gov. JRA-55 data was downloaded from ftp://ds.data.jma.go.jp. ERA-Interim data was downloaded from http://apps.ecmwf.int/datasets/data/interim-full-moda/levtype=sfc/.
- Balmaseda M et al (2015) The ocean reanalyses intercomparison project (ORA-IP). J Oper Oceanogr 8:s80–s97Google Scholar
- Behringer D, Xue Y (2004) Evaluation of the global ocean data assimilation system at NCEP: the Pacific Ocean, paper presented at eighth symposium on integrated observing and assimilation systems for atmosphere, ocean, and land surface, Am. Meteorol. Soc., SeattleGoogle Scholar
- Cronin MF, Fairall CW, McPhaden MJ (2006) An assessment of buoy-derived and numerical weather prediction surface heat fluxes in the tropical. Pac J Geophys Res 111:C6Google Scholar
- Dukhovskoy DS, Bourassa MA, Petersen GN, Steffen J (2017) Comparison of the ocean surface vector winds from atmospheric reanalysis and scatterometer-based wind products over the Nordic Seas and the northern North Atlantic and their application for ocean modeling. J Geophys Res 122:1943–1973CrossRefGoogle Scholar
- Zhang L, Kumar A, Wang W (2012) Influence of changes in observations on precipitation: a case study for the climate forecast system reanalysis (CFSR). J Geophys Res 117:D8Google Scholar