The impact of resolution and formulation on model simulations of an east coast low

Australian east coast lows are severe weather events that involve synoptic and sub-synoptic-scale processes. As such, it is of interest to examine the sensitivity of numerical simulations both to model formulation, and to model resolution. In this study, two models were used: a version of the current operational regional model (RASP) of the Australian Bureau of Meteorology that contains an improved representation of the boundary-layer processes; and the non-hydrostatic mesoscale model of the UK Meteorological Office (UKMO) which was designed specifically for very high resolution prediction. The RASP model was run at three different horizontal resolutions: 150 km (the current operational resolution), 50 km and 15 km. The UKMO model was run only at 15 km, which is within its design range. The east coast low of 31July-2 August 1990 was chosen for the sensitivity study. Large gains in accuracy were obtained in the RASP model simulations as the resolution was increased. This was true of all key variables including mean sea level pressure, low-level wind field strength and structure, and rainfall amounts and distribution. However, the RASP model simulations at 15 km were beginning to show some signs that the convection scheme was not appropriate for such high resolutions. The UKMO model simulation was generally of similar quality to the RASP model forecasts, but was slightly superior in the depiction of rain-band shape and coastal wind values. In summary, it is clear from the simulations that increasing the resolution of RASP leads to improved forecasts of Australian east coast lows. However, an equally important finding is that the improvement cannot go on indefinitely as at very high resolutions the representation of some physical processes in the model should be designed specifically for use at these high resolutions.