Home | About Us | Contacts | Help | Feedback | SEARCH Global | Australia | NSW | Vic. | Qld | WA | SA | Tas. | ACT | NT | Ant. | Weather & Warnings | Hydrology | Climate | Numerical Prediction | About Services | Learn About Meteorology | Registered User Services |

Global Guide
Main Menu

Global Guide to Tropical Cyclone Forecasting:
CHAPTER 4: THE HABITATION LAYER

4.4 SURFACE WINDS AT LANDFALL

4.4.1 Horizontal Asymmetry

It has been known since the very earliest analyses (Reid, 1838) that the cyclone wind field over the ocean is asymmetric with a maximum generally to the right (left) of the direction of motion in the Northern (Southern) Hemisphere. Although there are exceptions to this rule (Shapiro, 1983), the available observations are generally so poor as to negate any detailed analysis, especially under operational conditions. The general forecast rule, therefore, is to assume that the above asymmetry, with an amplitude given by adding and subtracting the tropical cyclone translational speed. The surface wind can readily be estimated by taking 0.7-0.8 of the gradient wind (Powell, 1982).

A combined empirical and theoretical study using the Holland (1980) wind and pressure profiles (Section 8.7) by Hubbert et al. (1992) found that the best combination was to locate the maximum wind slightly in the forward quadrant at an angle of 70o to the direction of motion. The amplitude of the asymmetry was given by adding the tropical cyclone translational motion to the symmetric wind field given by the derived pressure field. A constant reduction of 0.7 from the gradient level to the surface was used, together with a frictional turning of 20o (Shea and Gray, 1973). The analysis is automated with input of central pressure, radius of maximum winds and past two positions only required. Independent testing by G. Foley (personal communication, 1990) indicated a close fit to available anemometer records in a wide range of tropical cyclones, so that this method is recommended for use in estimating the horizontal wind structure within a couple of hundred kilometres of the centre of tropical cyclones.

Additional asymmetries over the ocean arise from asymmetric convection and the development of low-level jets. Convective asymmetries can be associated with increased inflow and with strengthened winds, usually on the inward side (Holland, 1987). When local stabilisation of the boundary layer occurs (eg from ocean cooling by mixing and upwelling) a local inertial acceleration and development of a low-level jet may occur (Moss, 1978; Anthes and Chang, 1978; Holland, 1987). There are no objective methods of incorporating this information into operational analyses, but forecasters should be aware of the potential.

At larger distances from the core, synoptic influences dominate in determining the overall wind asymmetry. For example, a tropical cyclone cradled by a strong subtropical high, may result in strong winds extending hundreds of kilometres from the cyclone. Such synoptic asymmetries are within the resolution of most observing platforms.

An objective technique for estimating the mean radius of 30 kt winds has been developed at JTWC by Martin and Holland (JTWC, 1992). This technique uses the Dvorak intensity estimate to estimate the parameter b in the Holland equations (Section 8.7) and the overcast cloud shield size (determined by the -65oC TBB isotherm) to estimate rm. Asymmetry is based on the cyclone movement and latitude. Operational use over several years has indicated that the technique provides useful analysis guidance.

As a tropical cyclone approaches land, the maximum winds tend to be located over the water just off the coast, with winds just inshore being around 20% less (Powell, 1982; Tuleya et al., 1984; Jones, 1986). When a significant mountain range is present, the coastal confluence and maximum wind region may develop whilst the cyclone is a considerable distance off shore (Holland, 1984c,d). In some numerical modelling studies (Jones, 1986), a new convective ring cycle has been initiated, leading to a temporary weakening of the cyclone. This could be followed by a rapid increase in surface winds as the new eye-wall contracts, following the process described by Willoughby et al. (1982).

Figure 4.13: Twelve hour decay rate as a function of central pressure for tropical cyclones in the USA (dots) and from the numerical study by Tuleya et al. (1984, triangles).

All tropical cyclones weaken after landfall at a rate that is approximately proportional to the intensity at landfall (Fig. 4.13). There is some evidence also that larger tropical cyclones tend to decay slower than do small systems. Immediately following landfall, the maximum wind belt tends to expand outwards and the surface pressure outside the core region may drop slightly (Miller, 1964; Tuleya and Kurihara, 1978). Presence of significant mountains causes more rapid disruption and weakening of the cyclone, but may also be associated with transient development of local jets (Tuleya and Kurihara, 1978; Padya, 1978).

4.4.2 Wind Variability and Gustiness

The local variability and gustiness of the winds in a tropical cyclone will be considerable and will be unpredictable in any detail. As a general rule, the surface wind speed will be higher over the ocean than over the land, but the gustiness will be higher over land and especially in mountainous terrain. This is illustrated by the anemometer traces in Fig. 4.14. The oceanic site of Willis Island (Fig. 4.14a) and the flat terrain of Onslow (Fig. 4.14b) result in much lower gustiness than does the mountainous terrain of Cairns (Fig. 4.14c) and Mont Desert Alma (Fig. 4.14d).

The ratio of maximum gusts (defined by the peak 2-s wind) to mean 1- and 10-min winds at an elevation of 10 m is indicated in Table 4.2. These provide both a single estimate and a potential range. Also shown in Table 4.2 are the ratios of peak gust over various terrains to that over the ocean; this includes the expected reduction in mean winds over the ocean and provides a ready estimate of the likely peak gusts for surface winds of a tropical cyclone approaching the coast.

Several dynamical factors also may cause substantial variations in the surface winds. Mountains and other obstructions can produce local jets, rotors and other microscale variations in the winds. The cooling of the surface air over flat land may result in a stabilisation of the boundary layer and substantial reduction in mean surface winds. Local, downburst-type severe wind-storms may then occur when upper flow mixes down to the surface (Fujita, personal communication, 1980). Recent analyses by Fujita (personal communication, 1992) of Hurricane Andrew in Miami also indicates that very small, tornadic-like circulations may have occurred in the surface flow and produced short-lived but extreme winds. As documented by Novlan and Gray (1974,) tornadic disturbances also can occur in large numbers in the outer circulation of landfalling tropical cyclones.

Figure 4.14: Anemometer traces for tropical cyclone conditions over differing terrain: a) the low-lying island of Willis Island off the eastern Australian coast, 6 February 1957; b) the flat coastal terrain site of Onslow in Western Australia, 4 March 1958; c) Cairns, which lies on the Australian east coast with mountains up to 1600 m immediately behind, 6 March 1956; and c) Mont Desert Alma at an elevation of 420 m in Mauritius, 6 February 1975 (from Bureau of Meteorology, 1978; after Whittingham, 1964 and Padya, 1976).

Contents Chapter 5