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 2: TROPICAL CYCLONE STRUCTURE

2.1 INTRODUCTION

Tropical cyclone structure forecasting has received considerably less attention than has motion forecasting. This is also a more difficult problem since the dynamics are more highly nonlinear and occur on smaller scales under conditions that make them very difficult to observe. Analysis and forecasting techniques are therefore highly empirical. Although high resolution numerical models may provide useful forecasts in the future, empirical methods are expected to provide the backbone of the forecasting effort. Research in the next decade should improve our understanding and conceptual models of structure change processes.

The forecast aids presented here are based upon our conceptual understanding as of 1990. Because of the lack of published material in this area, many of the check sheets have been developed specifically for the Guide and in some cases they have not been rigorously evaluated under operational conditions. Improvements will depend heavily on operational use, critical testing and exchange of information and ideas between the forecast and research communities.

2.1.1 General Principles

As with any type of weather forecasting, predicting tropical cyclone structure involves analysing the present conditions and estimating how they will change, i.e. monitoring and forecasting. In the absence of aircraft reconnaissance information, monitoring tropical cyclone structure is usually a problem of inferring intensity, size, and rainfall from frequent satellite imagery and reconciling the estimated structure with infrequent in situ observations from land stations, ships, and rawinsondes. Aircraft reconnaissance data simplify the analysis, but still leave considerable difficulties. Structure change forecasts are made using short-term indications from satellite cloud signatures and longer-term indications from synoptic conditions. These indications are based on a blend of forecaster experience and conceptual models of tropical cyclone dynamics. Because of the essentially subjective nature of this process, each section in this chapter is accompanied by a check-list to guide the interpretation and introduce a maximum amount of objectivity and consistency between forecasters.

Recommendation: Because structure changes are poorly understood and difficult to forecast, it is especially important that forecasters:

1. keep the needs, vulnerability, and response capacity of the community in mind;

2. know the range of forecast errors and communicate these as necessary; and

3. keep a record of all forecasts, the reasoning that went into them, and the results.

Tabulations of responses to all of the check-list items and the future structural changes of the tropical cyclone being forecast, especially if maintained on a computer database, would allow the rules to be clearly evaluated and refined after only a few seasons.

Expert systems provide an attractive way of logging this information, and of utilising it in subsequent system upgrades. Some experimentation is currently occurring with the use of expert systems; for example, prototype systems are being developed for use in the Australian Tropical cyclone Workstation (F. Woodcock, personal communication, 1993). However, since such systems are not yet in widespread use, no detailed mention is made here.

2.1.2 Notes on Interpreting Satellite Imagery

Accuracy of Satellite Methods: The almost universal method of estimating intensity from satellite imagery follows Dvorak (1984). This approach provides an orderliness and common approach to analysing satellite data for intensity and intensity trend and is the single most successful techniques used for cyclone intensity. However, used alone, it also is one of the most dangerous methodologies. Because it manages to keep orderliness and a "climatological" approach, it also deceives unacceptably when explosive deepening occurs.

A further concern is the lack of appreciation amongst forecasters of the inherent variability in satellite analysed intensity (Sheets and Grieman, 1975; Sheets and Holland, 1980). Because the Dvorak analysis is empirical, it has been "calibrated" based on aircraft reconnaissance in the North Atlantic and western North Pacific only. Other basin relationships between Dvorak CI (Current Intensity) Numbers and cyclone intensity (measured as lowest sea-level pressure or highest sustained winds) either assume the Atlantic or Pacific scale or else have been modified locally using rare direct observations. A particular concern here is the different wind-averaging times used in different ocean basins (see Section 1.3.3).

Recommendation: Cyclone intensity estimates based on the Dvorak technique have significant variability and potential for error, which is often overlooked in basins without other direct intensity information. We recommend that forecasters take great care to incorporate the real level of uncertainty into their forecast policy.

Diurnal Variability: Tropical convective systems, especially those at sea, undergo large diurnal variations because of the differences in the shortwave and longwave radiation budgets of clear and cloudy regions (McBride and Gray, 1980). Tropical cyclone structure changes, on the other hand, show little substantiated diurnal preference. Because many forecasting methods use trends in satellite imagery, it is critical that forecasters be aware that cloudiness, and therefore the satellite signatures of tropical systems, also show a marked diurnal variation unrelated to structural changes. Short-term trends in satellite imagery can be very misleading and cause major forecast errors.

The diurnal variability of satellite-observed cloudiness has been described for the Atlantic (Browner et al., 1977), the Australian region (Lajoie and Butterworth, 1984), and the western North Pacific (Zehr, 1987). The times of maximum and minimum cloud area in a satellite image depend upon the cloud type and therefore the range of IR (InfraRed) brightness temperatures (T_b) being considered:

1. The strongest diurnal variation is in very cold (T_b < -65^oC) clouds associated with deep convection. The maximum area of such cold cloud occurs between 0300-0600 LST (Local Solar Time) and the minimum between 1200-1800 LST.

2. The area covered by cirrus clouds (-15^oC < T_b < -45^oC) also follows a strong diurnal cycle but with an areal maximum around 1800 LST, at the time of the minimum in deep convective cloud area.

3. The diurnal cycle in total cloudiness (T_b < -15^oC) is weaker and follows that of cirrus, with a late afternoon (1800 LST) maximum.

4. Heavy rain follows the same diurnal cycle as convection. The maximum occurs between local midnight and dawn over the ocean, but can be anytime over land, especially when orographic effects dominate.

 

An idealised example of cloudiness variation is shown in Fig. 2.1. Because of these diurnal cycles, a tropical disturbance or cyclone often appears to be intensifying between local midnight and dawn and weakening in the afternoon. It is best to compare the current satellite picture with one from 24 h ago to identify trends associated with structure changes. Reaction to 6-12-h trends should only occur when they appear to be going against the expected diurnal cycle. For example, a reduced diurnal oscillation between morning and afternoon may indicate that the tropical cyclone is intensifying, and a sudden decrease in the area of very cold cloud may indicate that intensification has stopped temporarily. These non-diurnal trends are covered in Section 2.3.2.

Contents Chapter2.2