The sensitivity of landfalling tropical cyclone tracks and intensity to Indian Ocean SST anomalies

The impact of anomalies in east Indian Ocean SSTs on tropical cyclone activity is becoming increasingly recognised in terms of its landfall consequences for northwestern Australia In the 1998/99 tropical cyclone season, four tropical cyclones (TCs) crossed the West Australi an coastline over a two month period. This was a period of large SST anomalies over the east Indian Ocean. The four storms were TCs Billy (December 1998), Thelma (December 1998), Vance (March 1999), and Gwenda (April 1999). TC Thelma was the first category 5 (most intense) ever known to have developed in the Timor Sea . Estimated peak gusts reached 320 km/hr as it crossed the northwest coast of Australia. TC Vance devastated the town of Exmouth as it crossed the coast as a category 4 TC. Gwenda reached category 4 before weakening as it crossed the coast. Billy crossed the coast after weakening from a category 2 earlier in its development. During that period the Indian Ocean dipole was strongly developed. The Indian Ocean dipole refers to a sea surface temperature pattern that features positive anomalies in the northeast (over Indonesia and off the northwest coast of Australia) and negative anomalies in the southwest Indian Ocean. This pattern persisted throughout the 1998/99 tropical cyclone season. In this study, the sensitivity of TC motion and intensity to the dipole is investigated using the adjoint of the UNSW HIRES numerical weather prediction (NWP) model for the four tropical cyclones. Also the role of the SST anomalies alone, is investigated. The results obtained in this study should be expected to apply to other tropical cyclone basins.

It has been recognised for some time now that NWP model sensitivity studies can be carried out using the adjoint of the NWP model. Adjoint methods can be used to identify regions and variables that have the largest impact on the model forecasts. In such cases, small perturbations grow rapidly and quantitative estimates of this sensitivity can be made. Here, we examine the effects of SSTs in the east Indian ocean, using the adjoint method. Expressed simply, gradients in the cost function or scalar forecast aspect, J, are calculated. J is a function of the model variables, X . The initial and forecast values of X are defined as X0 and Xf . The (non-linear) response of J to changes in X0 provide the sensitivity we are seeking. The perturbed initial state is in this case one which contains small changes in the SST field.

Results obtained using the adjoint method were focussed on TCs nearing landfall. They show that the (positive) SST anomalies themselves are more important than the dipole and that they have a significant impact in increasing cyclogenesis, with more intense TCs. The TC track forecasts are also found to be very sensitive to areas of high SST anomalies. The adjoint approach was used to show sensitivity, nearing landfall, to other variables such as the middle and upper tropospheric wind patterns, especially wind shear. The most dramatic case was a 24 hour weakening of up to 50 hPa in the case of TC Gwenda that can be identified a priori, with the adjoint sensitivity approach.