The impact of FGGE data coverage and improved numerical techniques in numerical weather prediction in the Australian region

This paper describes the results from an extensive series of experiments aimed at enhancing the skill of the Australian Numerical Meteorology Research Centre (ANMRC) limited area analysis/forecast system. Significant increases in model performance were achieved, in the sense of reducing S1, skill scores and in forecasting precipitation, in three ways: (1) improved data base since the beginning of FGGE; (2) the application of a more realistic lateral boundary specification, particularly nesting within a hemispheric forecast model; and (3) the reduction of truncation errors by either decreasing the grid spacing or (more efficiently) employing higher order difference operators. The improvement in skill is sufficiently large to suggest that when implemented operationally the quality of routine prediction of synoptic scale patterns over the Australian region will be considerably improved.