Revisions during 1991

12 February 1991 A change was introduced in the analysis so that the departures of the observations are now calculated against a first-guess valid at the time of the observation, obtained by interpolation from first-guess at 3, 6 or 9 hours range. Therefore, asynoptic observations are used more consistently.

9 April 1991 The following modifications were implemented in the forecast model cycle 38:

1. Modifications to the cloud and radiation subroutines, both to improve the cloud/radiation processes and to facilitate their use with increased vertical resolution. In particular, the overlapping of cloud layers was changed from a random overlap to a maximum overlap assumption.



2. A revised scheme for vertical diffusion in the free atmosphere, developed following the switching off of vertical diffusion above the PBL in 1988; the vertical mixing represented by the revised scheme is smaller by an order of magnitude than under the scheme which was operational until 1987.



3. Two revisions to the convection scheme, to ensure consistent cloud physics within the cloud model, and to introduce a cloudtop temperature check to improve the onset determination of shower precipitation.

1 May 1991 A new procedure for the quality control of satellite temperature profiles was implemented; a pre-selection of the data to be passed to the analysis is made, using the high resolution cloud-cleared radiances received from NESDIS.

17 September 1991 A high resolution analysis and forecasting system at T213 31 levels was put into operations as cycle 39. In addition to the change of resolution, the new model includes the following aspects:

• the Gaussian grid used for the computation of the physics is now a reduced Gaussian grid, i.e. the number of grid points along a latitude circle decreases towards the poles in order that the spacing of the points be approximately 60 km on the whole globe. The effective resolution in the free atmosphere is around 100 km. at half wavelength;



• several changes are made to the dynamics of the model, including use of a semi-Lagrangian advection scheme and modified horizontal diffusion. A change in the link between the dynamics and the parametrisation of vertical diffusion is expected to reduce the surface wind by about 3% over sea and about 10-15% over land;



• a new version of the cloud scheme, mainly to allow the stratiform clouds to form in any number of layers instead of the three layers in the previous version.

An article from the ECMWF Newsletter describing these changes is given in pdf (380KB) file.

26 November 1991 A technical error in the model library was corrected in cycle 40. It was caused by an insufficiently discriminating table look-up in the long-wave radiation code, which created an erroneous systematic cooling between 150 and 400 hPa and warming above.

10 December 1991 Enhancements to the quality control of AIREP and SATOB data were introduced. In particular, the restrictions to the use of SATOB data over land have been relaxed.