|Title||Global stratospheric temperature bias and other stratospheric aspects of ERA5 and ERA5.1|
|Publication Type||Technical memorandum|
|Secondary Title||ECMWF Technical Memoranda|
|Authors||Simmons, A, Soci, C, Nicolas, J, Bell, B, Berrisford, P, Dragani, R, Flemming, J, Haimberger, L, Healy, S, Hersbach, H, Horányi, A, Inness, A, Munoz-Sabater, J, Radu, R, Schepers, D|
The ERA5 analyses of lower stratospheric temperature exhibit a pronounced cold bias for the years from 2000 to 2006. This is due to specifying background error covariances for the data assimilation that were inappropriate prior to availability during 2006 of GNSS radio occultation data in sufficient numbers to constrain a cold bias of the assimilating ERA5 model. A new set of analyses, termed ERA5.1, has thus been produced for the period from 2000 to 2006 using the background error covariances that were used to produce the ERA5 analyses for the years 1979 to 1999. ERA5.1 also includes the more restrictive ensemble assimilation of SBUV ozone data that was used in production of ERA5 for 1979 to 1999.
ERA5.1 provides analyses with better global-mean temperatures in the stratosphere and uppermost troposphere than provided by ERA5. ERA5.1 stands up well in comparison with ERA-Interim and other reanalyses in the lower stratosphere, although there are also lower-stratospheric temperature differences between ERA5 and other reanalyses in the 1980s and 1990s. These are due in part to differences in radiosonde temperature bias adjustment. The pronounced near-tropopause cold bias of ERA5 from 2000 to 2006 has implications for the representation of stratospheric humidity, for which ERA5.1 performs better, though by no means perfectly. ERA5.1 does not exhibit the spuriously high values of ozone that occur close to the South Pole in the polar nights of 2003 and 2004 in the ERA5 analyses. Synoptic evolution in the extratropical stratosphere is seen to be very similar in two cases involving splitting of the stratospheric polar vortex and secondary vortex formation by dynamical instability. ERA5.1’s representation of the QBO agrees slightly better with radiosonde wind data than that of ERA5. The dataset formed by merging ERA5.1 with ERA5 is generally more homogeneous over time than ERA5 alone. It nevertheless is problematic for global-mean upper stratospheric temperature for all but the most recent ten or so years.
ERA5.1 is very close to ERA5 in the lower and middle troposphere.