On 14 March 2006 a number of improvements in convection and boundary layer physics were implemented in the operational global model. These changes were aimed at improving tropical performance and deliver better tropical winds and significant improvements in the vertical structure of temperature and relative humidity. Precipitation is also beneficially reduced over tropical oceans and increased over tropical land. Consistent improvements were also seen in extra-tropics. Overall the physics package alone improved the global NWP index by between 2%-3% with biggest impacts in June-August.
At the same time improvements were made to the use of satellite data. AIRS data has been re-introduced, with an earlier error corrected and more data used. This gives benefits both in terms of NWP index and in the representation of moisture data. It also means we are now able to fully realise the benefit of the improved vertical resolution in the global model (50 levels introduced in December 2005) through improved satellite data processing. The satellite package also includes an upgrade from Meteosat-7 to Meteosat-8 satwinds, with the latter showing improved quality. ERS-2 scatterometer winds were also enabled for the North Atlantic providing an opportunity to capture significant synoptic events.
The combined impact of physics and satellite upgrades in a recent parallel trial was a 4% improvement in the global NWP index which is one of the biggest improvements in performance in recent years.
At the same time, the data assimilation system for the North Atlantic European (NAE) Model was upgraded from '3DVAR' to '4DVAR'. This follows a 15-month project to build on a similar development in the global model introduced in late 2004. The Met Office is one of only 2 operational centres worldwide to run 4DVAR in a regional configuration. 4DVAR makes more advanced use of the forecast model within the mathematical procedure for fitting the latest observations. It is able to make more effective use of observations distributed in time.
Results of trials in different seasons showed an improvement of between 2% and 5% in the UK Index which measures the skill of forecasts of near-surface weather. Errors in mean sea-level pressure and 10-metre wind are reduced. Forecasts of rainfall and cloud are improved.