Tuesday, 25 January 2011
Typhoon Morakot (2009) brought record-breaking rainfall over southern Taiwan, with maximum 96-h accumulated rainfall of 2,874 mm on the slope of the Central Mountain Range from August 6 to 10, 2009. Practical predictability of such extreme rainfall is limited by uncertainties in both the initial states and the forecast models. In this paper, we present results from a high-resolution (4-km) ensemble based on the Advanced Research WRF (ARW) model, initialized with the ECMWF 0.225º×0.225º operational analysis with perturbations compatible with analysis uncertainties added. To gain insights on the impact of precipitation parameterization, we perform six sets of ensemble experiments with different combination of cumulus parameterization and explicit microphysics on the 4-km and 12-km meshes, each with the same 8-member ensemble initial states. The results indicate that the model simulations are very sensitive to the choices of cumulus parameterization schemes for the 2-way interactive, triple nested meshes with horizontal resolution of 4, 12, and 36-km. Prior to landfall, all the model storms propagate in a westward direction without significant variation. However, when the storms start to interact with the Taiwan topography, they behave drastically differently between the two clusters of experiments with the two different cumulus parameterization schemes, Kain-Fritsch (KF) scheme and Betts-Miller-Janic (BMJ) scheme. The ensemble members with the use of KF scheme tend to drift northward along the northeast seashore of Taiwan and some even without making landfall. In sharp contrast, most of the model storms with the use of BMJ scheme pass through the Central Mountain Range of Taiwan, and move further westward or northward. The variability in storm track then results in significant variability in rainfall amount and distribution. Further analysis of the results indicates that the storm track variability is related to the variability in storm structure and intensity, with the use of different cumulus parameterization schemes. These results suggest that precipitation parameterization is an important source of model uncertainties, and they need to be properly taken into consideration for the design of a mesoscale ensemble system. This is especially important for the prediction of extreme rainfall events such as Typhoon Morakot (2009).
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner