5.3                                   Additional aids to analysis over the ocean

One additional difference with respect to the Southern Ocean and the Antarctic region is that their vast expanse has long been recognised as the most significant data void in the world. Within that region the South Pacific Ocean sector qualifies as the most deficient of all (Streten and Zillman, 1984). Increasingly satellite–derived numerical estimates of sea surface wind speed and direction and of atmospheric–layer thicknesses, as well as drifting buoy data, are assisting in overcoming the shortfall. However, these data are not always available and so the specific meteorological analysis techniques that have evolved to deal with the data sparsity of the seventies and eighties still have relevance.

The Junker (1977), and Guymer (1978) analysis techniques discussed below offer sound methods of analysis when conventional upper–air and ATOVS/TOVS data are not available in a given time frame. They also provide good training for an analyst in the identification of the key meteorological factors in a satellite image.

5.3.1                                Techniques for estimating MSLP  

Although numerical models are indispensable in forecasting cyclones, the sparsity of meteorological data over ocean regions may cause problems in accurate analysis and lead to incorrect forecasts especially in the Antarctic. It is helpful, therefore, to identify cyclones at an early stage of development, using satellite imagery to estimate the core or central pressure associated with their cloud patterns. King and Turner (1997, p. 190–193) provide a succinct overview of the Streten–Troup (1973) method for such estimation.

Similarly, Junker (1977) has developed a technique of cloud–based interpretation following the studies of Rogers and Sherr (1965). Pictures with characteristic cloud patterns of developing or intensifying cyclones were compared with the lowest observed pressure. They are grouped into six groups of 10–hPa increments between 1000 and 960 hPa with the typical cloud band structure.

When a wave starts to develop, the cloud band becomes bright with an anticyclonic bulge and core pressure of more than 1000 hPa. With increasing development, the pressure falls below 990 hPa and a dry slot begins to form on the rear edge of the solid bright clouds. If the intensification continues, the pressure falls to 980 hPa while the cloud band wraps itself nearly 3/4 of the way around the cyclone centre. If the central pressure reaches 970 hPa, the cloud band has completely surrounded the centre, as seen in Figure 5.3.1.1 with the mean diameter (MD) of the low.

Below 970 hPa, the cloud band wraps itself 1–1/2 times around the centre while a circular cloud pattern around the centre indicates still lower pressure.

For the data–sparse regions of the southern Atlantic between South Africa and South America and also during expeditions in the polar North Atlantic, a regression analysis has been carried out by the Deutscher Wetterdienst (DWD) (German Meteorological Service) maritime meteorology department, which relates parameters of the cloud band with typical core pressures of intensive cyclones.

As Figure 5.3.1.2 shows, a typical cyclonic cloud band was associated to the distance from core position to the first inner cloud band, called outer radius (OR), to the diameter of the core (ID) and to the core pressure that could be measured sometimes by buoy data or by observation through the German ship "Polarstern" near the centre of the low. Figure 5.3.1.3 is the MSLP analysis corresponding to the satellite image.

The relation between central pressure and the cloud structure parameters in the sub–Antarctic waters of the South Atlantic during the season from October to December was found as:

P = 0.0683 * OR – 0.172932 * ID – 0.03752 * PHI + 974.76231                   Equation 5.3.1.1

Where:

P   = pressure in hPa;

PHI          = geographical latitude (without sign);

OR           = “outer radius” in km;

ID = “inner diameter” in km.

The results are in good accordance with the results of Junker (1997), with errors in pressure estimates of less than +/– 10 hPa. Using the pressure value 12 hours before at the location of the low–pressure centre instead of the constant value of 974.76231 yields even better results.

A similar classification of various stages of cyclonic development by characteristic cloud patterns was made by Guymer (1978). The structure and the centre of the vortex are arranged relative to the anomaly pattern of each type of the cyclone stage. These anomaly patterns were developed and used over a 10–year period at the World Weather Centre in Melbourne. One of the decision rules is that the MSLP anomaly, which refers to the seasonal mean pressure for a regularly shaped vortex in mid–latitudes, is 1.5 times its east–west diameter.

5.3.2                                Techniques for upper–air analysis  

Although the period since the IGY has been one of steady improvement in the availability of observations and the quality of analyses for the Southern Hemisphere, the lack of upper–air observations is still a problem. Several investigations (e.g.: Trenberth, 1979; van Loon and Shea, 1988; Karoly, 1989; Pook, 1992) have shown that the data are satisfactory for the analysis of large–scale features, but the paucity of observations from conventional observing systems has been an impediment to more detailed analysis. In situations where only limited upper–air observations are available alternative analysis techniques can be employed to determine atmospheric thickness. Fronts and low–pressure systems located close to the Antarctic may arguably not always, or even often, exhibit "classical" features and so the models discussed here may have less applicability near the Antarctic coast than at lower latitudes.

Guymer (1978) developed a system of analysis whereby an experienced analyst could combine interpretation of satellite imagery with mean thickness and MSLP fields to construct a 1000–500–hPa thickness chart. Using the USA’s Environmental Science Services Administration (ESSA) satellite visible–light (VIS) imagery and extending the pioneering work of Martin (1968), Zillman (1969), Zillman and Price (1972 ), Troup and Streten (1972), Streten and Troup (1973) and Streten and Kellas (1973), he produced thickness anomalies that could be tied to recognisable features of the cloud field. With the increased availability of TOVS data, this method has fallen out of favour but the technique can still provide a semi‑objective means of employing satellite imagery (IR and VIS) to complete an analysis and to assist with pseudo observations for numerical models (Guymer, 1978, his Appendix). It is also instructive for analysts operating in remote environments such as Antarctica to use the technique as a means of improving their interpretation of satellite imagery.

In order to apply the “Guymer Technique” it is necessary to adopt a realistic model of Southern Ocean depressions. The main features of the model employed by Guymer (1978) are shown in Figure 5.3.2.1. Guymer (1978) found that the frontal cloud shown as (D) in Figure 5.3.2.1 is a region of above average 1000–500–hPa geopotential thickness whereas the region on the western flank of the vortex (E) is characterised by air with 1000–500–hPa thickness well below average values for a given latitude and time of year. The critical change from above–average to below–average thickness is delineated by a 'line of zero departure" that is shown in Figure 5.3.2.2. The thickness ridge (point of maximum positive departure) and thickness trough (region of greatest negative departure) are also shown in Figure 5.3.2.2.

Once the 1000–500 hPa thickness field was established, then by simple addition to the 1000 hPa field (derived from the mean sea level field through the hypsometric equation) then the 500–hPa field can be defined, as can the 850 and 700 hPa levels using regression (Guymer 1978, page 2 of his Appendix).

A more recent model of a typical extra–tropical depression has been given by Bell et al. (1988) in which they argue that the position of the upper tropospheric wind maximum or jet stream can be inferred from the characteristic cloud pattern that occurs on the warm (equatorial) side of a baroclinic zone. Bell et al. (1988) identify the three principal components of this model as the jet associated cloud, the vorticity comma cloud, and the vortex deformation cloud.

Guymer (1978) has argued that the maximum–thickness gradient, indicating the approximate position of the polar front jet stream, will be found between 536 and 544 dam when the curvature of the contours is cyclonic, and between 536 and 528 dam when the curvature is anticyclonic, the latter case resulting in a super geostrophic wind. Similarly, Gibson (1989) has used thickness criteria to identify a higher–latitude wind maximum that he calls the Antarctic Jet.

5.3.2.1                          Climatologies

The construction of a realistic thickness or MSLP anomaly field depends on the choice of climatological data sets. Initially, average fields were available from the work of Taljaard et al. (1969) and Taljaard (1972) but later climatologies have been developed from the Australian Bureau of Meteorology Southern Hemisphere Analyses (Le Marshall et al., 1985), the NCEP reanalysis programme and the ECMWF reanalysis programme.

5.4                                   Analysis over the interior 

Preparing meaningful analyses over the continent is a difficult task. Apart from the fact that there is very little in the way of data upon which to base the analysis, almost the entire continent is above 2,000 m (~6,500 ft), with significant portions rising over 4,000 m (~13,000 ft). So as mentioned earlier (Section 2.4.6 and Section 2.6.1) reducing pressures reported from the AWSs to a mean sea level value is simply not a valid thing to do given the extent of the altitude reduction and the problems in defining a mean virtual temperature for the layer from the surface down through several kilometres of ice. Because of this, the surface layer of the atmosphere over the continent requires a different form of analysis; one that accurately and meaningfully defines the surface layer and one that can be merged with the more standard MSLP analysis performed over the oceanic and coastal regions. Three differing approaches have been used.

The first of these is to use a standard reduction technique and simply analyse MSLP over the continent. This is still performed in some analysis centres but the product does not provide any useful meteorological information. The observed wind field is certainly not geostrophic and as such the flow defined by the MSLP chart highly inaccurate. The second technique involves performing a geopotential height analysis at the first standard pressure level that does not intersect the continent. The 500–hPa surface meets this requirement and has been a standard level analysed in Antarctica for some time. Unfortunately, it suffers some of the same problems as the MSLP field in that the inferred wind regime from the analysis bears little resemblance to the surface wind field defined by the AWS data. Strong decoupling between the surface layer and the free atmosphere can make this form of analysis difficult to interpret and the amount of data provided from upper–air reporting stations is even more limited than the amount of AWS data available.