Tropical Cyclone Forecasts with a High Resolution ACCESS Model

Tuesday, 19 April 2016
Plaza Grand Ballroom (The Condado Hilton Plaza)
Xingbao Wang, Bureau of Meteorology, Melbourne, Australia; and J. D. Kepert, N. E. Davidson, and Y. Xiao

The offshore oil and gas industry in Australia's northwest operates in a harsh and remote environment, in which tropical cyclones pose a significant risk to safe and efficient operations. Improved forecasts offer a significant opportunity to mitigate this risk, and to this end we are developing a new tropical cyclone NWP system for the northwest. The new system builds on our experience with the existing operational system ACCESS-TC, but differs from it in being on a larger and fixed domain instead of relocatable, by running twice-daily instead of only when a cyclone has formed, by having higher resolution (4 km grid instead of 11 km), by forecasting to 5 instead of 3 days, and by including a one-way coupled wave model. The system is called ACCESS-TCX, with the X standing for “extended”, and aims to address specific issues around forecast length, tropical cyclogenesis and wave prediction. We have tested various ACCESS model configurations for this purpose over both northern Australia and the northwest Pacific Ocean, nested in either the global or regional ACCESS models. Intensity forecasts are substantially better than our existing tropical cyclone model, due to the combination of higher resolution and 4D-Var initialisation. Track forecasts are presently slightly worse, which may indicate that further tuning and calibration of the assimilation is required. We have also found that the system performance is sensitive to model resolution, physics, initial condition and lateral boundary conditions. The introduction of a new dynamical core to the model illustrated forecast sensitivity to the model dynamics, with the newer dynamical core producing stronger TCs. While we have demonstrated that ACCESS-TCX has the potential to provide good TC track and intensity forecasts, development continues to determine the best combination of model dynamics, physics, and data assimilation.
- Indicates paper has been withdrawn from meeting
- Submission entered in competition