Known quality issues with ERA-Interim data

Please see our publications and reports for detailed discussion of many aspects of ERA-Interim data quality.

Spurious shifts in ERA-Interim time-series related to changes in the observing system include:

• Shifts in global mean precipitation during the period January 1992 - December 2009, due to the assimilation of rain-affected radiances from SSM/I. See Section 5.2.1 in Dee et al. 2011 for details.
• A discontinuity in upper-stratospheric temperatures (at levels 5hPa and higher) associated with the introduction of radiance data from AMSU-A in August 1998. See Section 8 in Dee and Uppala 2008 for details.
• Warming of the lower stratosphere by approximately 0.2K in December 2006, with the introduction of GPS radio occultation data from the COSMIC constellation, which partly corrects an otherwise unconstrained cold bias in the assimilating model. See Poli et al. 2010.
• Slight excess warming of upper-tropospheric temperatures due to the assimilation of growing numbers of warm-biased temperature measurements from aircraft, beginning in 1999 (Dee and Uppala 2009). After December 2006, this drift is somewhat reduced with the introduction of GPS radio occultation data from the COSMIC constellation (Poli et al. 2010).
• The homogeneity of the ozone time-series is affected by the availability of observations from different satellite ozone instruments, which fluctuates over time. See Section 4.5.1 in Dee et al. 2011.

The low cloud cover (LCC) in ERA-Interim is underestimated if there is both cloud in the profile either side of the the medium cloud cover (MCC)/LCC boundary and the cloud fraction is less in the lower layer than in the higher layer. This problem also applies to MCC for the high cloud cover (HCC)/MCC interface.

Prescribed sea-ice fraction and sea-surface temperature data for ERA-Interim are as for ERA-40 prior to January 2002; i.e. based on the NCEP 2DVAR dataset to the end of June 2001 and the NCEP OISST V2 product from July to December 2001. From 1 January 2002 ERA-Interim follows the ECMWF operational forecasting system, using the NCEP RTG product until 31 January 2009, and the OSTIA product from 1 February 2009.

Additional issues related to sea ice:

• ERA-Interim used corrupted sea-ice cover data near Antarctica on 1 August 2005.
• Between 1 January 1989 and 31 January 2009, prescribed sea-ice fraction data for ERA-Interim is always 1 for locations northern of 82.5 degrees latitude North.
• Between 1 January 1979 and 31 December 1988, and after 1 February 2009, prescribed sea-ice fraction data for ERA-Interim found to be less than 0.2 are set to 0.

ERA-Interim snow analyses from 1 July 2003 to 23 February 2010 are affected by a geo-location error introduced during the processing of NESDIS snow cover data. Data locations were shifted by about 100km toward the South-East, causing incorrect removal of snow in some coastal areas in the Northern Hemisphere during winter. This error has been present in the ECMWF operational forecasting system from 18 March 2004 until 23 February 2010, when it was discovered and corrected (26 May 2010).

Data for February 2009 were replaced in the MARS archive on 15 May 2009 1500UTC, and on the Data Server on 18 May 2009 1200UTC (1 July 2009).

The values of convective available potential energy (CAPE) at step 3 hours are erroneously zero in the CAPE products (13 May 2009).

Due to an error in the ECMWF forecast model, incident solar radiation at the top of the atmosphere is approximately 3Wm-2 higher than intended. This has resulted in a slightly warmer (approximately 1K) upper stratosphere for the entire ERA-Interim period (4 May 2009).