Evaluation of Parametric Tropical Cyclone Surface Winds over the Eastern Australian Region

When assessing tropical cyclone hazards, users are presented with several simplified parametric models to describe the surface wind field of tropical cyclones. These parametric models are used routinely for risk assessment of cyclonic winds, as well as for input to surge and wave models used in coastal hazard assessments. Differences between the models include the formulation of the parametric cyclone model, the way winds above the boundary layer are specified at the surface and along-track parameters that describe the cyclones’ size and shape. Hazard studies based on thousands of synthetic tropical cyclone (TC) events require a validated model representation of the surface wind field. Here, we assess three different TC parametric vortex models with input from four along-track parameter studies of the TC size and shape, based on statistical formulation of the relationships to observed TC intensity, geographic location, and forward transition speed. The 12 model combinations are compared to in-situ 10-minute observed surface mean wind speeds for ten TCs that made landfall over Queensland, Australia, which occurred over the period 2006 to 2017. Empirical wind reduction factors to reduce gradient winds to the surface are recalculated for the more recent TCs at both offshore (ocean, small islands, reefs, and moorings) and on onshore (land) locations. To improve the wind comparisons over ocean and land a secondary reduction factor was developed based on an inland decay function. Pearson correlations for the unadjusted modelled peak wind speed from 118 instances of a TC passing a weather station sit between a range of 0.57 and 0.65 for the 12 model combinations. Using the secondary reduction factor based on the inland decay function increases the range of correlation to 0.74 to 0.81. Based on the assessment of the instances of peak surface wind speed correlations, bias and root mean square error, along with the correlation 48-hours around the peak, the top ranked performing model combination for the region was an along-track parameter study with a double vortex model, both previously tested for the South Pacific basin.

The regression plots in the top row show model validation and correction of peak 10-minute surface mean wind speed. Each point represents the peak wind recorded/modelled at a station for a single TC instance, circles are onshore, and crosses are offshore stations. The application of the wind reduction factor over land brings more of the onshore points in line with the observations (top right). The wind field plots in the bottom row show the effect of the wind reduction factor over land for cyclone Yasi at the time of landfall in February 2011 (land is to the left each plot).