Some techniques for high resolution prediction of tropical cyclones (TC s) will be described, and results presented for a number of events in the Australian region. The vortex-generation, assimilation, initialisation and forecast components may be summarised as follows.
1. Based on observational and theoretical studies of TC structure and motion, synthetic data are generated at any desired resolution to define a storm s circulation, consistent with observed location, size, intensity and past motion. The method involves : (i) determination of the environmental flow by filtering of the symmetric and asymmetric components of the TC circulation in the (old) objective analysis, (ii) generation of a new symmetric vortex, which is merged at outer radii with the old, relocated symmetric vortex, and (iii) construction of vortex asymmetries by requiring that the observed motion equal environmental steering plus the asymmetric flow.
2. At first a 12 hour collection of the standard observational data (satellite and conventional), is assimilated at coarse resolution, without the synthetic data. The analyses thus formed define the large scale environment of the storm and are used in the generation of synthetic data. The 12 hour assimilation, using the additional artificial data, is then rerun and a 48 hour forecast made at coarse resolution. The objective analyses from this assimilation are used as first guesses for high resolution analyses of the storm and its environment. In this case the synthetic data are generated at sufficient resolution (20k) to reproduce in the objective analysis the observed central pressure and winds at the radius of maximum wind.
3. Initialisation for fine mesh prediction consists of 24 hours of diabatic, dynamical nudging through 6 hourly, high resolution analyses. During this phase the vorticity and surface pressure fields are preserved, while infrared satellite cloud imagery is used to reconstruct the moisture and vertical motion field to be consistent with the observed distribution of cloudiness. This procedure initialises and balances the vortex, and can reduce erratic track, intensity and structure behaviour during the early hours of the forecast.
4. The prediction model is a high resolution version of the operational limited area model of the Australian Bureau of Meteorology, and includes high order numerics and advanced physical parameterisations including a mass flux convection scheme. The fine mesh forecast is one-way nested in the coarse mesh prediction.
The encouraging quality of the forecasts has been demonstrated in numerous case studies of TC events, including some well-documented real time, forecast failures. For the 17 events considered and even with 2 rather poor forecasts, mean forecast track errors at 24 and 48 hours are 115 and 259 kms. These are very competitive with errors in official and climatology-persistence forecasts. Given the uncertainties in observed central pressure estimates, forecasts of intensity are also quite encouraging. General discussion will focus on system idiosyncrasies and some remaining, unresolved issues. By conference time, we hope to have early results from real time trials in the western Pacific.