A set of observing system experiments (OSE) has been run to measure the impact of various observation types in the assimilation and forecast system of the Canadian Meteorological Centre (CMC). The observing systems tested were ATOVS (AMSU-A) radiances, SATOB atmospheric motion winds, HUMSAT (humidity estimates from satellite), aircraft, radiosondes and surface observations. The reference system is the CMC operational 3D variational assimilation system. The latest version of the 3D-Var analysis was implemented in December 2001 with improvements to assimilate temperature and surface pressure instead of geopotential, and to the number and type of observations assimilated. This study currently covers the period from 17 December 2001 to 27 January 2002. Results from a summer period should be ready by the time of the symposium. The results are grouped in three series of OSEs for satellite, conventional and aircraft observations over the following areas: Northern Hemisphere, Southern Hemisphere, Tropics and North America.

Experiments done by denying sets of conventional data confirm the usefulness of these observations in all areas. In the Southern Hemisphere and Tropics areas, the satellite data clearly dominates and is the main contributor to the NWP forecast performance. The experiment done with no satellite data (ATOVS, SATOB, HUMSAT) indicates an interaction between ATOVS and SATOBs data, since the overall impact if often more important than for the total of ATOVS and SATOBs impacts taken separately. Over North America the use of aircraft observations provides a definitive improvements in the quality of CMC global assimilation and forecast system. The results show a clear positive impact from using temperature and wind data from aircraft observations, but the contribution of the aircraft winds is somewhat more important than the contribution of aircraft temperature.