ECMWF 6-month report (DEMETER project)

Francisco Doblas-Reyes

The DEMETER hindcast integrations are steadily running and following closely the ERA40 initial data availability. At the time of writing this report (September 2001) two models are running their experiments: ECMWF and MeteoFrance. They have completed all hindcasts in the period from November 1986 to May 1992. This means that there are almost 6 years of hindcasts (23 start dates). The rest of the models will begin their integrations before the end of the year. Depending on the ERA40 data   availability we expect a steady growing of the ECMWF DEMETER hindcast data set.  The two available models show a very different pattern of precipitation bias in summer over the Asian-Australian monsoon area (ERA40 data have been used as the reference climatology). The same applies to the lower and upper level winds, with no model being the best as a whole. Possibly linked to the model bias is the low skill of monsoon rainfall. For instance, both models seem to be unable to properly simulate the precipitation differences over India in 1987 and 1988, despite the significant ENSO skill showed by the NINO3 and NINO4 indices.   Nevertheless, dynamical indices of the circulation over the northern Indian ocean and southern Asia show some hindcast skill of the atmospheric circulation, though the length of the sample avoids the assessment of the statistical significance.