From ERA-40 to the 3rd generation ECMWF reanalysis ERA-Interim

Sakari Uppala


Since the second ECMWF reanalysis ERA-40 a new reanalysis has been started. The data assimilation has been configured to use 12 hour 4D-Var, T255L60 assimilating model, new humidity analysis, upgraded model physics and variational bias correction of radiance data. This new reanalysis system started from January 1989 and will produce an interim reanalysis (ERA-Interim) for the data-rich 1990s and 2000s, and will be continued as an ECMWF Climate Data Assimilation System (ECDAS) until superseded by a new extended reanalysis. The analysis has now reached 1999 and is expected to catch near real-time in the latter part of 2008.

Several of the problems experienced in ERA-40 have been eliminated or significantly reduced in ERA-Interim: most notably a too-strong tropical oceanic precipitation that was marked from the early 1990s onwards and a too-strong Brewer-Dobson circulation in the stratosphere. Precipitation, which is higher in both ERA-Interim and ERA-40 than in the Global Precipitation Climatology Project (GPCP) over the tropical oceans, is closer to GPCP in ERA-Interim. ERA-40 and ERA-Interim are nevertheless in closer agreement with each other than either is to the GPCP estimate. A further indication of improvement of the hydrological cycle in ERA-Interim comes from diagnosis of the global balance of precipitation and evaporation. Comparison of the accuracy of 10-day forecasts from ERA-Interim and ERA-40 with those from operations provides further evidence of the improvement of forecasting systems over the years.

The results highlight the fact that instead of being viewed as largely independent "one-off" exercises, reanalysis has come to be seen more as an iterative process. Developments in modelling, data-analysis techniques and computing power together with data rescue efforts and experience from other reanalyses will produce a succession of reanalyses with increasing quality.