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

2.3 TROPICAL CYCLONE INTENSITY

The Dvorak (1984) analysis is the worldwide standard for tropical cyclone intensity monitoring in the absence of aircraft reconnaissance and is the most common method of intensity forecasting as well. An important part of the technique is a climatological development rate, which provides a basis for estimating intensity changes.

Research by Holliday and Thompson (1979) indicates that 75% of all western North Pacific tropical cyclones deeper than 920 hPa have experienced a period of rapid intensification of 42 hPa d^-1 or more. Extreme deepening rates of nearly 100 hPa d^-1 have been observed. All tropical cyclones, even the weaker ones, should therefore be regarded as potentially serious.

One conceptual model of tropical cyclones at sea is that they are self-amplifying systems. They will intensify until they reach a Maximum Potential Intensity (MPI) unless their surroundings disrupt them, as is frequently (and fortunately) the case. The potential intensity is primarily a function of Sea Surface Temperature (SST) and tropopause temperature (Emanuel, 1988b), so passage over colder water (or land) reduces the MPI. Strong vertical shear of the environmental flow is the most common factor limiting intensification in tropical and subtropical latitudes at sea. Tropical cyclones with a compact core of maximum winds and strongest convection are thought to intensify more rapidly, as are those that are well below their potential intensity. Another commonly held view is that interactions with upper-level troughs, either of tropical or subtropical nature, may further tropical cyclone intensification under the right conditions.

The threat of rapid intensification resulting in a very destructive tropical cyclone should therefore be considered greatest for compact, well-organised circulations with warm SST, a high tropopause, and relatively low vertical shear of the environmental flow. The Dvorak method contains detailed procedures for evaluating the satellite signature of a tropical cyclone in terms of its current and near-future intensity and is included in abbreviated form in Section 2.3.2 for convenience. Work sheets designed to modify the Dvorak forecast development rate based on other information are then presented and discussed. Also described are other forecast methods that show promise but are not fully developed or require specific regional implementation.

2.3.1 Observing Methods

Direct measurements of intensity in the form of wind and pressure observations are seldom available. The eye and area of maximum winds cover a very small area and are unlikely to affect a station directly, especially for a ship whose captain is intent on avoiding the opportunity to observe the most severe part of the tropical cyclone. Observations from anywhere within the circulation are helpful (see Section 2.4) but alone reveal little about intensity (Holland, 1981c; Weatherford and Gray, 1988). The area of destructive winds can be very concentrated, especially in the case of a rapidly developing tropical cyclone.

The most common estimates of intensity are those inferred from satellite imagery using Dvorak analysis (Section 2.3.2). It is also possible to monitor the upper-tropospheric warm anomaly directly using passive microwave observations from satellites. Thermodynamic sounding retrievals from the NOAA Microwave Sounding Unit (MSU) have been statistically related to central pressure reduction and maximum winds for tropical cyclones in the North Atlantic (Velden, 1989) and western North Pacific (Velden et al., 1991). The technique has not been used operationally but is expected to have errors similar to the Dvorak analysis. It also performs less effectively on rapidly developing tropical cyclones, but does have the advantage that each estimate is independent of earlier analyses. Thus the Velden technique does not suffer from an accumulation of errors that may occur with a Dvorak analysis.

2.3.2 The Dvorak Intensity Analysis Technique

The Dvorak technique forms the basic intensity analysis method for most tropical cyclone forecast offices (McBride and Holland, 1987). It exists in various forms and has been modified by many forecast offices for their own local procedures. The version used here has been abstracted from Dvorak (1984) and follows that used in WMO (1987). The overall approach is summarised in Figs. 2.4-2.6 and described below.

Figure 2.4a: The Dvorak IR analysis check sheet.

Figure 2.4b: The Dvorak IR analysis check sheet (ctd).

Figure 2.5a: The Dvorak VIS analysis check sheet.

Figure 2.5b: The Dvorak VIS analysis check sheet (ctd).

Figure 2.6: Work sheet for applying the Dvorak technique.

Step 1: Locate the Cloud System Centre (CSC): The CSC is the focal point of all curved lines or bands of the cloud system, or the geometric centre of the eye, if present. A full description of its location is provided in Section 3.2.3.

Step 1a: Initial Development: Disturbances showing signs of development into tropical cyclone intensity (see Section 2.2) are classified as T1. To be so classified, the disturbance must have the following three properties:

ID1. It has persisted for 12 h or more;

ID2. It has a cloud system centre defined within an area having a diameter of 2.5^o lat or less, which has persisted for 6 h;

ID3. It has an area of dense, cold (<-31^oC) overcast greater than 1.5^o lat in extent that appears less than 2^o lat from the centre. The overcast may also appear in cumulonimbus lines that curve around the centre.

Considerable variability of cloud pattern often is observed during the T1 stage. Curved cirrus lines often develop in association with upper-level anticyclonic shear and may indicate patterns far more advanced than T1 at the time of the initial classification. These patterns do not involve deep tropospheric circulations at this time and will be short lived so that the Data T-number on day 2 may be analysed as less than that on day 1, as is consistent with the two-stage conceptual model of tropical cyclogenesis. These weak systems are also especially prone to diurnal and non-periodic oscillations in convection which have no relationship to the slowly varying circulation (Section 2.2.1). The rule is to never lower the T- number at night during the first 24 h of development. Note that a classification of T1 forecasts tropical cyclone intensity (T2.5) 36 h after the T1 observation only when the environment is expected to remain favourable. A minus symbol is used after the T1 to indicate a T1 pattern that is not expected to develop.

Step 2: Determine the Pattern Type: The manner in which the cloud system centre is defined determines which of the following pattern types would be used. When the cloud pattern being analysed does not resemble one of the patterns, proceed to Step 3.

Step 2a. Curved Band Pattern

Step 2b. Shear Pattern

Step 2c. Eye Pattern

Step 2d. Central Dense Overcast (CDO) Pattern

Step 2e. Embedded Centre Pattern

General Analysis Rules:

GA1. When short-interval pictures are available, use the average measurement of all of the pictures with well-defined features taken within the 3-h period ending at analysis time;

GA2. When two or more T-number estimates are made from the same picture, use the estimate closest to the Model Expected T-number (MET);

GA3. When in doubt concerning ambiguous features, bias the analysis toward the MET;

GA4. When two or more clear-cut Data T-number estimates are made for the same picture, and there is uncertainty on which is the most representative, split the difference between the two.

Step 2a: Curved Band Pattern: The intensity estimate is derived by measuring the arc length of the curved band fitted to a 10^o logarithmic spiral overlay. The spiral overlay is fitted to the axis of the coldest overcast gray shade (most dense clouds) within the cloud band and should roughly parallel the overcast edge on the concave side of the band. When the band indicates two possible axes, use the one with tightest curvature. Cellular clouds and small breaks in the cloud band should be ignored. As the curved band pattern evolves it will usually be defined by the dark gray shade, but may at times appear defined in warmer or colder shades of gray. At times the boundaries of the band must be interpreted from its form in previous pictures.

The associated Data T-number (DT) is taken from Fig. 24a or 2.5a. Curved band measurements may be used with both visible and EIR pictures until an intensity of DT4.5 is reached. For EIR patterns greater than DT3.5 use measurements from Fig. 2.5a.

During the first 2 days of development (T1 to T2), the amount of overall band curvature may change excessively, very little, or even decrease somewhat for short periods. Thus, the tendency should be to raise the T-number by one during the first 24 h of development as long as the band remains curved enough for T2 and clear signs of weakening or rapid development are not apparent. It is also important to allow at least 24 h to pass between a T2 and a T4 classification. Even though the coiling process has been observed to be faster than this at times, the surface pressure does not fall accordingly.

Step 2b: Shear Patterns: Shear patterns most commonly occur during early development and weakening. They are identified by the cold clouds moving to one side of the cyclone and developing a sharp edge.

The DT is derived from both the method of defining the CSC and the distance between the low cloud centre and the dense, cold overcast. For tropical storm intensity (T2.5-3.5), the centre will be defined by parallel, circularly curved low cloud lines with a diameter of 1.5^o lat or less near or under the edge of a dense, cold (<-31^oC) overcast cloud mass. For weaker systems (T1.5-2.5), the low cloud centre will either be poorly defined in spiral lines within 1.25^o lat of the cold overcast clouds, or circularly defined near a small, dense overcast (< 1.5^o lat diameter).

Step 2c: Eye Pattern: Eye patterns are analysed for systems which have been at T2 or greater for 24 h and have an eye (Section 3.2.3). The data T-number is derived from,

where: E the Eye number, EA the Eye Adjustment number (Fig. 2.4a, 2.5a) and BF is the Banding Feature number (Fig. 2.5a). The eye adjustment factor is determined by using the table next to Step 2a in Fig. 2.4a and the rules at the top right hand corner of Fig. 2.5a. The BF addition is used with EIR pictures only when the T-number estimate without the BF is lower than the model expected T-number.

Step 2d: CDO Pattern (Visible imagery): CDO patterns consist of a dense, solid-looking mass of clouds covering the cloud system centre Generally, well defined CDOs of at least 1^o lat width are associated with tropical storm intensities whilst those measuring 2^o lat or more are associated with hurricanes. Banding features (BF) are usually added to the CF term for CDO patterns (Fig. 2.5a).

Step 2e: Embedded Centre Pattern (Infrared imagery): Embedded centre patterns are analysed when the tropical cyclone has had a previous history of a T3.5 or greater intensity and when the CSC is clearly indicated to be within a cold overcast (<9^oC). Curved cloud lines or bands within the cold overcast as well as the outer curved bands will indicate the location of the CSC within the overcast. The analysis of this pattern is similar to the eye pattern analysis except that no eye adjustment factor is added (Fig. 2.4a).

Step3: Central Cold Cover (CCC) Pattern: The CCC pattern consists of an approximately circular, cold or dense overcast covering the cyclone centre or comma head and obscuring the expected signs of pattern evolution. The CCC pattern usually is associated with arrested development (Figs. 2.4b, 2.5b). Care should be exercised under the following conditions:

CC1. Do not confuse a CCC pattern with a very cold comma pattern, usually indicated by a very cold comma tail and head, with smooth texture and an indication of a wedge; Curved cirrus lines or boundaries usually appear around the cold pattern and not around the CCC pattern.

CC2. Do not assume weakening in a CCC pattern when the comma tail begins to decrease in size, as is commonly observed; also the CCC often warms as the eye of the T4 pattern begins to develop.

Step 4: Past 24-h Trend: The trend of the past 24-h intensity change is determined by other observations, or by comparing the current and 24-h past cloud features. Development is generally associated with increased organisation and better defined central features. Signs of development (D) include:

D1. Curved band pattern: Curved band coils farther around the CSC;

D2. CDO pattern: CDO becomes larger or an increase in banding features is noted;

D3. Shear pattern: CSC becomes more tightly defined in curved cloud line or appears closer to the dense overcast.;

D4. Eye pattern: Eye is more embedded, more distinct (warmer), less ragged, or is surrounded by colder (smoother textures) clouds, or more banding features;

D5. No significant (non-diurnal) warming of the cloud system is noted.

The cyclone has weakened (W) when its cloud pattern indicates a persistent trend opposite to those listed in (1)-(5) above. Watch in particular for patterns that become sheared or exhibit warming (lowering) of cloud tops that is NOT associated with the diurnal cycle. The cyclone has become steady state (S) when:

W1. A central cold cover appears in a T3.5 or greater tropical cyclone or has persisted for more than 12 h in a weaker cyclone;

W2. The CSC's relationship to the cold clouds has not changed significantly;

W3. There are conflicting indications of both development and weakening.

Step 5: The Model Expected T-number (MET): The MET is determined by using the 24-h old T-number, the D, S, or W, decision in Step 4, and the past amount of intensity change of the cyclone. When the growth rate has not been established in the case of new developments or reversals in trend, assume a past rate of change of one T-number per day; generally:

Rapid or slow past rates of change are established when two consecutive analyses showing rapid or slow pattern evolution are observed at 6-h or more intervals, or when one observation of strong intensification or weakening is made (see Step 10).

Step 6: The Pattern T-number (PT): The pattern T-number is used primarily as an adjustment to the MET when an adjustment is indicated, as shown in Figs. 2.4b, 2.5b.

Steps 7,8: Determining the Final T-number: The final T-number is determined from the rules in Figs. 2.4b, 2.5b. Constraints are used to hold the T-number within acceptable bounds and avoid misinterpretations of marked changes.

Step 9: The Current Intensity Number: The CI number relates directly to the intensity of the tropical cyclone and is determined from the T-number by the rules in Figs. 2.4b, 2.5b. Table 2.1 indicates the empirical relationship with intensity.

After each intensity analysis, the previous analyses of the tropical cyclone should be reviewed for consistency. When an error was made in the previous day's analysis, correct the T-number to provide a more accurate model-expected intensity. This correction may at times alter the current intensity analysis.

2.3.3 Forecasting Intensity Changes

At present, intensity change forecasts are made using the Dvorak model development curve (Section 2.3.2 Step 5), statistical methods, and subjective assessment of satellite imagery and environmental conditions. The statistical methods are similar to those developed for predicting motion but are generally less effective because intensity trends can change so quickly. Basin-specific statistical blends of climatology and persistence have been developed using multiple regression (Jarvinen and Neumann, 1979, Merrill, 1987) and Markov chain methods (Leslie and Holland, 1991). A statistical method combining environmental parameters (vertical shear, upper-tropospheric fluxes of angular momentum, and circulation size) with climatological parameters and an empirical estimate of the potential maximum intensity has shown promise for Atlantic basin tropical cyclones (DeMaria and Kaplan, 1993). These statistical models are developed in the same way as those used to predict tropical cyclone motion. Forecasters interested in developing such techniques for local use should refer to the section on tropical cyclone motion, especially regarding CLIPER-type methods and references to the statistical procedures used.

Figure 2.7 provides a work sheet for intensity forecasting that merges the Dvorak method with other forecaster rules and incorporates some recent research results. It is an attempt to retain the useful and physically reasonable predictors from several sets of empirical rules and also to account explicitly for extratropical transition and transformation (Section 2.5). Many sets of empirical rules have been developed in the form of flow charts and check lists but few have been rigorously verified and evaluated⁽¹⁾. As noted at the end of Section 2.2, detailed records of the responses to all of the work sheet entries and the subsequent behaviour of the tropical cyclone would be extremely useful in refining this method.

The favourable and unfavourable satellite imagery signatures associated with cloud bands were developed in the Atlantic basin and their usefulness in the Southern Hemisphere has been questioned. They should be applied tentatively there and their usefulness documented further.

The vertical shear under environmental features should be calculated using area-averaged winds in the upper and lower troposphere. A method for computing such averages is presented in the appendix to Chapter 2.

The adjustment for "outflow jets" is based upon an Atlantic basin results (Merrill, 1988) that: 1. the outflow jet (Fig. 2.8) tends to wrap nearly all of the way around tropical cyclones that have stopped deepening well below their MPI; 2. the jet exits the region within a quadrant or two for intensifying systems.

The maximum potential intensity (MPI) can be estimated using the theoretical curve in Fig. 2.9 and maps of climatological or observed sea surface temperature (SST) and tropopause temperature, which should be close to the outflow temperature for more intense systems. The MPI so obtained is in terms of central pressure, which can be converted to a Dvorak CI-number for use at the end of the intensity change work sheet using Table 2.1 or a local CI/pressure relationship.

Figure 2.8: A method for defining the "principal outflow jet" using an upper-tropospheric streamline analysis. All streamlines which cross a circle around the tropical cyclone's core convective region are candidates, shown with intermediate line widths. Of these streamlines, the one which is associated with the strongest wind speed within 1000 km of the tropical cyclone center is designated the "principal outflow jet." In the example, it exits the storm at a bearing of 60° and crosses 1000 km radius at 150°, having passed through 90° of azimuth, or one complete quadrant.

Figure 2.9: Maximum potential intensity (central pressure in hPa) as a function of sea surface and outflow temperatures from Emanuel (1988b). An ambient pressure of 1013 hPa and a relative humidity of 80 percent are assumed.

1. See herbert (1978) for one example of a rigorous evaluation of a check sheet.

Contents Chapter 2.4