Validation of TC-LAPS structure forecasts of some significant 2004–2005 US hurricanes

Abstract

The Australian Bureau of Meterology's Tropical Cyclone Limited Area Prediction System (TC-LAPS) has been operational over the Australian Region and northwest Pacific since 1999. Forecast verification has been encouraging with long-term mean track and intensity errors at 48 hours of ~261km and 22hPa. Some remarkable forecasts of intensity have been obtained, but the quality from day-to-day tends to be inconsistent, suggesting that the system is capable of making skilful intensity forecasts, but the skill is often limited by errors in either initial vortex structure or the large-scale environment. Some examples will be shown at the meeting.

To validate TC-LAPS, we have made a limited number of forecasts for some significant US hurricanes, when the observational network is somewhat superior to that over the Australian region. The network appears to provide reliable analyses of the large-scale. The availability of ground truth for vortex structure, allows specification of TC vortices which are consistent with their environments.

TC-LAPS has 5 main components: 1. Data assimilation, to build from conventional observational data, the large-scale environment and outer structure of the storm. (Resolution is 0.3750 latitude-longitude on 29 sigma levels.). 2. TC vortex specification, to construct the inner-core, impose asymmetries consistent with the past motion, and to re-locate the circulation to its observed position. 3. High resolution objective analysis, to insert the specified TC circulation into its large-scale environment. 4. TC initialisation, to re-define the vertical motion field to be consistent with the satellite cloud imagery, build the secondary circulation, and initialise (balance) the vortex. 5. High-resolution prediction, to make the forecast with the generalised LAPS model, which includes advanced numerics and sophisticated parameterisations. (Resolution in steps 2 to 5 is 0.150 latitude-longitude on 29 sigma levels.)

For the limited number of forecasts made so far, results over the US have been encouraging. Figure 1. shows observed and forecast tracks and intensities for Hurricane Charley from a single base time. Somewhat similar quality forecasts for other storms suggest that TC-LAPS forecasts of track and intensity in the Australian Region may be mostly limited by the quality of initial conditions. Further validation of interesting track and structure forecasts will be presented at the meeting.

Figure 1. Left panel: Observed and 72-hour forecast tracks and intensities for Hurricane Charley from base time 12UTC, 11 August 2004. Table shows observed and forecast central pressures, and track errors (TERR) in kms. Right panel: Time series over 72-hours of estimated (green) and forecast (red) central pressure at 6-hourly intervals. Forecast max. wind is the red dashed curve.