6.6                                   Forecasting the main meteorological elements 

6.6.1                                Surface wind 

6.6.1.1                          Day to day wind forecasting: the range of considerations

Typically a forecaster is presented with the challenge of preparing a forecast for the coming day (or days) during which the wind might be forced (created) by one or more of a number of factors, for example, an approaching low–pressure system. The range of considerations that the forecaster will want to consider include:

·                         Departures from persistence: what mechanisms might operate during the forecast period to cause the wind to vary from what is happening now? Such departures would be based on the expected movement and development of synoptic and mesoscale pressure systems (and more particularly the associated pressure gradients; isobaric curvature; and isallobaric fields) and diurnal/thermal effects (for example: katabatic flow). The magnitude of the departures will also be of concern: will a strong wind event occur that might affect operations, or, conversely, will winds be light, or even calm?

·                         Prognoses and wind models: what prognoses and/or wind models (numerical and/or manual) are available and how good are they? Do they have any systematic biases? Here one would hope that studies on the reliability of numerical models (for example, Leonard et al. (1997) and Adams (1997) have been undertaken to assist in the assessment. Ad hoc assessments of numerical guidance reliability would of course be undertaken by the forecaster through comparisons of model output with analyses and observations during forecast operations.

·                         Local effects: how will the wind behaviour be affected by factors such as: orography (for example: barrier winds; lee rotor effects); and boundary layer considerations such as surface roughness; turbulence; and heat budget (net effect of insolation).

The factors mentioned above are not solely peculiar to the Antarctic environment and the forecaster will access a range of techniques that are applicable to mid–latitude forecasting. The sections below examine some key features of surface wind forecasting that might be considered in the Antarctic context.

6.6.1.2                          Diurnal/thermal wind systems 

Although most atmospheric motions (that is, wind) have as their primary cause differences in heat budgets, it is appropriate to consider wind circulations that are directly caused by diurnal or local thermal effects (for example: sea and snow breezes; inversion winds; and katabatic winds) separately from the larger‑scale atmospheric motions.

Sea and snow breezes

As mentioned in Section 6.5.1.3, sea breezes may occur in the Antarctic where exposed rock is adjacent to open water and it is probable that snow breeze circulations occur in areas where exposed rock are adjacent to snow or ice. While the editors are not aware of any established forecasting rules for sea or snow breezes in the Antarctic, intuition suggests that, in simple terms, where significant thermal contrasts are established then a wind circulation is likely to be set up. However, these breezes are only likely to be observed when the prevailing synoptic wind velocity is suitable. For example, for a sea breeze to occur at Hobart Airport, Australia, irrespective of the land–sea thermal contrast, the opposing surface synoptic wind component can usually be no more than about 11 m s^–1 (Whitehead, 1972).

References that might be of value in the sea/snow breeze forecasting problem for very high latitudes include: Johnson (1984); Lin and Stewart (1986), Segal and Arritt (1992); and Kozo (1982 a &b).

Katabatic winds: some definitions and clarifications in the Antarctic context

It is relevant that a clear terminology be established with respect to the Antarctic surface wind. This is important in view of the various definitions of katabatic flow that appear in the literature, and due to the more casual use of "katabatic" in the spoken word where the term is often used to refer to any sort of wind in the Antarctic. (In much the same way that the term "tidal wave" is incorrectly used to refer to a tsunami). Some examples of the various treatment of "katabatic" in the literature are as follows. Schwerdtfeger (1970, p. 290) reports an earlier suggestion by H. H. Lettau that a distinction should be made between the equilibrium flow ("inversion wind") that is possible over the relatively slightly inclined snow and ice slopes of the Antarctic interior and the apparently non–equilibrium downslope winds often experienced near the steep slopes of the continental margin. Schwerdtfeger (1970, p. 291–292) goes on to highlight some ambiguity regarding the term "katabatic", noting that one definition allows three different types of wind: "inversion wind"; "Foehn winds"; and "bora" type katabatic flow. Murphy and Simmonds (1993, p. 528), assume katabatic flow to be the deviation of the resultant wind from geostrophy, but acknowledge the difficulties of such a definition. (Phillpot (1997, p 63) reports a common definition of an Antarctic coastal katabatic wind flow being one that entails in part a speed decrease from “a fairly high surface value (~12–14 m s^–1 (~35–30 kt) to 5 m s^–1 (~10 kt) or less by about the 1 km or 850–hPa level”. The Meteorological Glossary clearly links the term “katabatic wind” with the downslope gravitational flow of air cooled by contact with ground that has lost heat through long wave radiation (Meteorological Office, 1991, p. 166).

Parish and Waight (1987) discuss the forcing of Antarctic katabatic winds. Through the assistance of a two–dimensional model these authors highlight some key differences between the characteristics of katabatic flow over the more gently orographically sloped interior when compared to the katabatic flow over the order of magnitude steeper orography near the coast. These authors show various graphs that indicate the difference, a brief summary of which is given in Table 6.6.1.2.1.

However, it is clear that both the inland downslope flow and the coastal downslope flow are "katabatic" according to the Meteorological Glossary definition and the differences are more ones of size and time to development rather than being intrinsically dissimilar. Nevertheless, it may be inferred from Parish's (1988) review of Antarctic surface winds that he subscribes to the defacto convention of, in cases where outward long–wave radiation (radiative flux divergence) from a sloping surface is the primary forcing mechanism, retaining the term "katabatic" for the coastal wind regime and "inversion winds" for the winds in the interior.

Table 6.6.1.2.1     Inferred differences between coastal, near coastal and continental katabatic flows. (From Parish and Waight, 1987 (mainly p. 2,224–25).)

Characteristic	Coast (steep slopes 10 –1 to 10 –2)	Near inland (intermediate)	Inland (gentle slopes 10 –3)
Strength of katabatic flow	Strongest	Closer to "Inland"	Weakest
Onset/cessation	Most abrupt	Closer to "Inland"	Most gradual
Direction	Most downslope (less balanced by Coriolis)	Closer to "Inland"	Across slope (more balanced by Coriolis)
Forcing	Unbalanced flow resulting from strong gravitational flow of radiatively cooled air (with resulting pressure gradients across the inversion around five times the inland case), opposed by friction to some extent, but Coriolis less important. (Turbulent exchange processes most important.)	Intermediate, probably closer to "Inland" characteristics	Near equilibrium balance between pressure gradient, Coriolis and friction

Now, in reality of course, there is usually a synoptic (or mesoscale) pressure gradient superimposed on the pressure gradient set up by the radiative flux divergence from the orographic slope. Phillpot (1997, p. 42–74) discusses surface wind behaviour in East Antarctica and refers to "enhanced katabatic" and "low level wind speed maximum" wind types (Phillpot (1997, p. 66–67 and 70–73)). The latter, at least, (Phillpot (1997, p. 660) clearly relates to "probable off–shore cyclonic activity". And, as is discussed in Section 6.6.1.3, Murphy and Simmonds (1993) relate how the synoptic pressure gradient can interact with katabatic flow to produce strong wind events.

So where does all this leave the Antarctic weather forecaster, who must, on the input side of formulating a wind forecast, understand the dynamics and thermodynamics (and related terminology) of the forces that drive the wind? Equally, however, on the output side, the forecaster must use language that portrays to the forecast recipient (client) an unambiguous forecast of wind velocity. In compiling this handbook the Editors take the view that forecasters should follow as close as possible to terminology that has developed after considerable thought by the research community provided that terminology has some meaning to the wider community.

The following terms are therefore adopted when referring to the Antarctic surface wind.

On the "input" or forecast formulation side:

·                         "Katabatic flow": we follow the Meteorological Glossary definition that relates to "downslope gravitational flow of air cooled by contact with ground that has lost heat through long wave radiation (Meteorological Office (1991, p. 166)).

·                         "Katabatic winds": we follow Parish and earlier authors (for example, Schwerdtfeger (1970, p. 290)) in thinking of these winds as being largely confined to the coast, and perhaps near coast, where the relatively steep slopes in the orography allow the "katabatic flow" to be strong;

·                         "Inversion winds": similarly, we follow Parish and earlier authors (for example, Schwerdtfeger (1970, p. 290)) in thinking of these winds as being largely confined to the Antarctic interior, where the relatively shallow slopes in the orography causes the "katabatic flow" to be weak;

·                         "Surface wind": we would recommend simply using the term "surface wind" in cases where non–katabatic forcing is a significant factor (for example, a strong synoptic pressure gradient). Here we would endorse the use of "blizzard" or "strong wind event" etc., provided that the term does not imply that "katabatic" forcing is the sole mechanism.

On the "output" or forecast dissemination side:

·                         "Katabatic flow or katabatics": are terms that most technical people would understand to be related to downslope flow in situations consistent with the "katabatic flow" definition above. (Meteorological Office (1991, p. 166). In other words, when dealing with the general community, "katabatic flow" or "katabatics", should be used in both the "katabatic wind" and "inversion wind" situations described above.

·                         "Surface wind": should be used if there is any doubt about the forcing mechanism or about whether the audience would misinterpret the intent of the forecast. Again simple terms such as "blizzard" or "strong wind event" may also be used if appropriate.

Winds over the interior – "Inversion winds"  

Parish (1988, p. 172) discusses the small contribution that synoptic scale maritime cyclones make to the wind climate of the Antarctic interior. The intrusion of maritime storms over the Antarctic continent does occur, however, (see, for example, Pook and Cowled (1997)), and a forecaster would need to use model guidance or experience built up from case studies to deal with these.

The prediction of the presumably more tractable "inversion wind" requires an assessment of the synoptic pressure gradient and the pressure gradient set up by the inversion set up by the sloping orography: from these the Coriolis force and frictional forces would be computed in some way to complete the approximate balance in the flow (see, for example, Parish and Bromwich, 1986). As may be inferred by Parish's Table 1 (Parish (1988, p. 172), wind direction for most inland stations has a high degree of constancy. For example, the directional constancy at Vostok is given by Parish as 0.91. The directional constancy at the Dôme C AWS is, however, much less at 0.51: this is to be expected given that this AWS, being at or near to the top of a dome would be in an area of almost no orographic slope and thus more prone to synoptic influences. However, for the most part the forecaster can use climatology to obtain at least a first guess of the inversion wind direction for many stations.

A better estimate of both direction and speed would come from an appropriate numerical model that incorporates the "inversion wind" thermodynamic and dynamic modelling. It would be the responsibility of relevant forecasters to push for the development of such models if they do not already exist operationally. Alternatively, model output statistic (MOS) type approaches, or pattern recognition – perfect prognosis approaches might also be developed.

Katabatic winds: near the Antarctic coast.

In this section we are dealing with winds generated by surface radiation flux divergence near the steep slopes of the Antarctic coastline. As mentioned, Parish and Waight (1987) discuss the forcing of Antarctic katabatic winds in general: in particular, the following is inferred as being useful in the forecasting of the katabatic flow near the coast.

·                         Extra strong or prolonged katabatic flow (of a NON–cyclonic storm type) might occur down–wind of a region where the inland "inversion" wind converges, thus providing a prolonged–nearly limitless supply of negatively buoyant air. (Parish and Waight, 1987, p. 2,226; Parish, 1984; Parish and Bromwich, 1987);

·                         Terrain slopes are in the order of 10 ^–2 to 10 ^–1  (Parish and Waight, 1987, p. 2,214);

·                         Strong radiational cooling exists (Parish and Waight, 1987, p. 2,225);

·                         Synoptic control (Gallée et al., 1996; Gallée and Pettré, 1998).

The first two points noted can be easily established by the forecaster by reference to the work of, for example, Parish (1984) and Parish and Bromwich (1986, 1987). (Figure 2.6.7.1.1 is taken from the last reference and shows idealised katabatic streamlines for average winter conditions). The third point (radiational cooling) requires the forecaster to predict radiational losses that result from clear skies, usually under anti–cyclonic conditions. It would be hoped that numerical modelling, either through direct physical representation, or through model output statistic methods, would ultimately be the best forecasting tool for the coastal katabatic. However, until such models become readily available, the forecaster will need to rely on the conceptual model approach implied here or on case studies peculiar to each site.

The fourth point is partially covered in the comments on radiational cooling. However, Gallée and Pettré (1998) suggest that synoptic control has an important bearing on the cessation of katabatic flow. Through numerical modelling these authors suggest that the pooling of cold air seaward of the coast causes a pressure gradient back towards the continent. As the radiational cooling weakens or ceases during the day, this pressure gradient may cause a reversal in the wind flow from katabatic to anabatic.

It may be inferred from the work of Gallée et al. (1998) that if the synoptic (or mesoscale) flow is weak then the surface warming is sufficient for generating an additional upslope buoyancy force (an effect no doubt enhanced by any sea breeze effects due to warming of exposed rock), and anabatic flow develops over the ice sheet in the afternoon. On the other hand, forecasters might note that when the large–scale wind is moderate and downslope, the piling up of cold air is enhanced and this has a dramatic impact on the flow. A sharp spatial transition is generated between downslope and the static pressure winds (antitriptic winds blowing towards the coast) over the ocean. This discontinuity moves toward the ice sheet interior in the morning and is responsible for the sudden cessation of the katabatic flow seen by static observers.

6.6.1.3                          Strong Wind Events (SWEs) 

Strong wind events are probably the most important winds events to be forecast in the Antarctic, as they can impact significantly on Antarctic operations and put lives at risk in some situations. Given that for blizzard conditions to exist, requires a wind speed equalling or exceeding gale–force (17.5 m s^–1 (34 kt)) then that speed is used here to give the lower bound to the 10–minute average wind speed in a SWE. There are potentially several categories of SWE:

·                         the "katabatic wind" in which the prime forcing mechanism results in the "bora" type wind (for discussion see Section 6.6.1.2);

·                         the "barrier wind" (for discussion see Section 6.6.1.4);

·                         downslope winds accelerated by gravity waves (for discussion see Section 6.6.1.5 below )

·                         hydraulic jumps (for discussion see Section 6.6.13);

·                         events in which synoptic (or mesoscale) weather systems play an important role, with or without a contribution from katabatic flow. The last category is the focus of this current section..

It may be inferred from Streten (1968, p. 52), when he refers to the work of Astapenko (1964), that the role of strengthening easterly pressure gradients between oceanic lows and continental anticyclones in coastal winds storms affecting East Antarctica was discussed at least as early as 1964. In more recent times, Murphy and Simmonds (1993) use a general circulation model to examine the role that such synoptic systems might have in SWE. The key conclusions from this recent work are that strong pressure gradients and strong katabatic flows can operate together to produce SWE, with each component in the order of three times stronger (in winter at least) than average. Moreover, it may be inferred from the work of Murphy and Simmonds that the role of the inland anticyclone is two–fold: not only does it assist with the development of above–average pressure gradients it also provides the subsidence and clear skies that are conducive to the development of the katabatic flow (Murphy and Simmonds (1993, p. 533).

Figure 6.6.1.3.1 is a NOAA composite IR image showing, in particular, a deep low north of Casey Station and clear skies in the ridge over the continent to the south, at about the time of a SWE at the station. The surface winds at Casey were above gale force for less than about 12 hours in this event, due in part to the mobile nature of the low. As discussed in Section 6.3, a long wave ridge east of a station is often instrumental in blocking the eastward movement of lows resulting in SWEs along the Antarctic coast south of the low (see also (Murphy and Simmonds (1993, p. 533). On 27 December the long wave pattern was conducive to progression of the short–wave features and so the SWE event at Casey was short lived.

Thus for predicting synoptically driven SWEs the forecaster will want to take into account, using, in particular, numerical model guidance and satellite data:

·                         pressure gradients: are they significantly (say around three times (Murphy and Simmonds, 1993) above average?

·                         has a high–pressure system allowed for radiative losses from the surface over the last few days to allow the formation of inversions and katabatic flow?

·                         local influences on the wind: for example, for stations on the north of the Antarctic Peninsula it may be enough just to consider the effects of the low, whereas, stations in the lee of ice domes may be affected by downslope flow due to gravity waves or lee waves (see Section 6.6.1.5).

SWEs forced by mesoscale lows need also to be considered in some circumstances. The forecast decisions will include whether such lows will form in the first place and, if so, the strength of the isobaric circulation.

6.6.1.4                          Barrier winds 

As discussed in Section 2.6.7.3, barrier winds are formed by synoptically driven boundary layer air with strong static stability impinging on high orography. In the case of the Ross Ice Shelf–Transantarctic Mountains’ situation, for example, with a synoptic easterly geostrophic wind, the depth of the cold air boundary layer increases from east to west, that is towards the mountains. A geostrophic balance is then set up between the Coriolis force and the pressure gradient force generated by the relatively high–pressure area in the cold air dammed against the mountains and the relatively low pressure further east over the ice shelf. Thus south to southwest winds, often quite strong, are generated below the height of the orographic barrier.

The forecaster therefore has to concern himself or herself with:

·                         is the orography conducive to the barrier wind effect: that is to say, is there high orography adjacent to a relatively flat area?

·                         is the synoptic situation conducive to a prolonged (actual minimum period not known?) piling up of boundary layer air with high static stability against the high orography?

·                         the magnitude of the air pressure differential set up between the barrier and the nearby flat surface;

·                         will local orography affect the barrier wind flow in some way, by either blocking it or channelling it?

Figure 6.6.1.3.1      A NOAA IR composite image for 27 December 1991 (the most western segment of the image was taken nominally at 1529UTC 27 December) showing, in particular, a low–pressure system north of Casey, and clear skies to the south at the time of a strong wind event at the station.

6.6.1.5                          Downslope winds: lee wave/gravity wave acceleration and other thermal effects  

Adams (1996) investigated a SWE at Casey Station and suggested that an internal atmospheric gravity wave was generated in the flow over Law Dome heralding a switch from separated flow over the Dome to a downslope wind. The strong gradient generated by the passage of a cyclone close to Casey provided the forcing for the wave over the dome, with the downslope wind, and induced gravity wave, producing the accelerated flow in the Casey area. Turner et al. (2001) investigated a SWE that occurred at Casey in March 1992 and have found that the SWE evolved in a somewhat similar manner to the case presented by Adams (1996).

The March 1992 case examined by Turner et al. (2001) is an example of a synoptically‑induced event that had a major impact on one sector of the Antarctic coastal region, causing great disruption to logistical operations and creating extremely dangerous conditions at Casey Station. It is arguable that the downslope flow mechanism discussed above contributes to SWE at other sites, although further work is needed to verify this speculation. Some prima facie supporting inferences come from Streten (1968) who examined the distribution of the duration of SWE about pressure minima for Mawson, Davis and Wilkes (now Casey), arguing that if Ball’s (1960) model of katabatic wind–synoptic pressure gradient interaction was realistic then there should be a bias towards a strengthening of winds after the pressure minimum, that is, when the coastal low–pressure system had moved east of a station. Streten found that for Mawson such a bias did exist but for Wilkes the strengthening of winds was skewed towards falling pressure, the data for Davis were too limited for a conclusion to be reached. Bromwich (1989, p. 693) points out that a similar increase in wind on a falling bar had been noted at Port Martin. Callaghan (1984) has updated Streten’s work for Davis and found that for 47 storms examined at Davis, 61.7% of the SWE commencements were associated with rising pressure (or just prior to the pressure minimum and on the rising bar); 27.7% were associated with falling and rising barometric pressure; and 10.6% commenced on a falling bar.

It is possible to argue that while each site has its own local geographic peculiarities, the essential ingredient in the itinerant coastal wind storms of the Antarctic is the downslope nature of the flow as distinct from the radiation–katabatic flow per se. Here one might make the distinction between the long–lived concentrated outflow regions (such as the well known Cape Denison–Commonwealth Bay area (Parish and Wendler (1991)) or katabatic flows clearly observed in glacial valleys such as mentioned earlier in the Vanderford Glacier near Casey, and the major storms that modulate the average condition along the Antarctic coast. This would be a generalisation from the diagnosis of the March 1992 storm discussed by Turner et al. (2001) that the strengthening of the synoptic pressure gradient identified by early workers (for example Streten (1968)) is the catalyst in bringing a variety of site dependent factors into play, but significantly, with each factor a contributor to negative buoyancy driven downslope winds. In most coastal and near interior areas the entrainment of radiationally‑turbulent heat flux cooled air will be important in providing the density differences, at least early in the storm. Some sites, such as Casey, which are at the bottom of a steep moraine slope, might find that early in a storm's evolution the strong surface winds override the station, or are separated from the upstream laminar flow by hydraulic jumps or lee eddies. While other sites, such as Davis, might have a similar situation compounded by the increased friction of intervening rough orography such as the Vestfold Hills (Targett (1998)). After a while, the invasion of air of maritime origin allows the radiative and turbulent heat flux contrast to be replaced by turbulent heat losses to the ice surface from the air from warmer regions to the north. In some cases, for example Casey, the energetics of the storm are added to by gravity wave motion causing downstream low to middle level warming of the atmosphere.

Even the bias to wind storms beginning at Port Martin on a falling bar might have a contribution from this source in view of the ridge–like orography east of the port. Moreover while there is a bias for wind increases at Mawson and Davis Stations to occur on a rising bar, there are still many SWEs that commence at these stations on a falling bar and we note the presence of significant orographic ridges east of these sites.

6.6.1.6                          Light winds

While there are probably no specific forecasting techniques available for the forecasting of light winds, the topic is briefly raised here simply to draw attention to the utility of such predictions. To maximise the safety of the most operations and to maximise the comfort of most people in the Antarctic (in particular to minimise the wind chill) it is usually preferable for outdoor activities to be conducted in light wind/calm situations.

In most cases the forecaster will simply be concerned with predicting situations with small pressure gradients and with reduced radiation losses to minimise katabatic effects. As noted in Section 2.6.7.3, light winds can also occur when strong wind streams are diverted around orographic barriers, which seems to be the case at Windless Bight.

In other situations care must be taken to interpret the synoptic/mesoscale situation carefully. For example, flow separation in the lee of orographic features can lead to light winds or reverse of flow just prior to a SWE. For example, in the 1991 situation depicted by Figure 6.6.1.3.1, the surface wind at Casey was southeast at 3.6 m s^–1 (7 kt) at 1500 UTC 27 December and by 2100 UTC 27 December the wind was easterly at 23 m s^–1 (45 kt). Following Adams (1996), it is likely that the strong winds passed over Casey (with rotor effects at the station) until the thermodynamics allowed the separation point to move downstream of the station and strong winds commenced. A similar sequence happened in the SWE described by Turner et al. (2001).

6.6.2                                Upper–level winds  

The Antarctic forecaster will have an interest in producing accurate forecasts of above surface tropospheric parameters for the following purposes:

·                         producing route forecasts for aviation;

·                         assessing synoptic and mesoscale developments;

·                         assessing the potential for precipitation or even just cloud (white–out).

There are few published studies that relate specifically to the analysis of the high–latitude troposphere above the surface and almost none that relate to how parameters such as upper winds might be forecast. Probably the most efficient methods for such forecasts are now–casting, using any available observational upper–level data, or the use of NWP output. An example of the latter is shown below. Figure 6.6.2.1 is a typical route taken by high–level Antarctic tourist flights (Figure 7.12.4.2.1) that depart Australia. Figure 6.6.2.2 is the NWP forecast cross–sectional wind, temperature, geopotential height and MSLP for the flight path shown in Figure 6.6.2.1.

In the absence of NWP data the forecaster might be tempted to produce his/her own upper‑level prognoses in which case the manual analysis techniques described in Section 5.3.2 might assist.

6.6.3                                Clouds  

The problem of forecasting cloud in Antarctica is similar to that in mid and low latitudes in the general approach to the problem. Namely, cloud masses need to be identified, air–mass movement assessed and the likelihood of development/decay considered. However, Antarctica introduces several problems not evident in other parts of the globe. Firstly, in the process of identifying cloud masses, low, or even middle level, cloud over the continent or sea ice may be all but invisible using standard IR or VIS channels from the AVHRR or GMS satellite systems. This problem arises from the cloud mass being at a very similar temperature to the underlying cold continental ice or sea ice surface giving rise to little or no discernible cloud signature in the IR image. Similarly, the visual characteristics of the cloud are almost identical to those of the underlying ice. The problem of assessing air–mass characteristics and movements at high latitudes is little different from other parts of the globe, other than the analysis process is weaker due to a lack of observational data at high latitudes and the numerical model guidance available is of a global nature and far coarser in resolution than available in other parts of the world. The general lack of observational data and reliance on low resolution models has a necessarily adverse impact on cloud forecasting.

6.6.3.1                          Cloud identification.  

The number of observational sites around Antarctica providing visual observations of cloud is so small that sole reliance on these sites to identify significant areas of cloud is pointless. Satellite imagery is almost the sole tool for cloud identification. Typically, Antarctic forecasting centres have at their disposal both geostationary satellite data and imagery from polar orbiting satellites. The geostationary data includes IR data at a nominal sub–satellite resolution of 4 km (but much coarser at the latitudes of Antarctica), and visual data with a sub–satellite resolution of 1 km (e.g. GMS5). Water–vapour imagery is also available, although at a coarser resolution. The primary polar–orbiting data is provided by the NOAA series of satellites. These platforms have five channels of AVHRR data comprising two transmitted IR channels, one channel a combination of transmitted and reflected IR, a visual channel and a near–visible channel. The sub–satellite resolution of these data are around 1 km.

Over the oceanic regions surrounding Antarctica the GMS IR and visual data provide a useful means of identifying cloud and differing air masses. However, by far the best source of data for the broad–scale identification are mosaics of the AVHRR IR images remapped into a polar stereographic projection at 4 km resolution. Images such as these are routinely produced at a number of Antarctic stations with mosaicing over six–hourly time steps. These images also allow the identification of intrusions of high level cloud over the continent such as occurs when broad–scale low–pressure systems move inland out of the polar trough.

For aviation and field party support where cloud has a significant impact on operations, small scale local regions of cloud are just as important as the broader scale cloud masses. The areas in which cloud impinges on operations include:

·                         areas of low cloud below alternate minima or safe operations for the aircraft;

·                         cloud producing icing problems for the aircraft or snow/precipitation resulting in a significant reduction in visibility;

·                         uniform decks of cloud causing white–out conditions. Over–plateau operations may be significantly effected by even cirrus if the high level cloud is uniform in nature.

The most effective means of identifying such cloud areas is through the use of the 1km resolution radiometry data from the NOAA polar orbiters. Each of the five channels on the NOAA has useful information for the identification of cloud.

High cloud is most easily identified using the IR channels (bands 4 and 5), as the cloud top temperatures associated with cirrus are typically cooler than the surface temperature of the continent. In cases where the cirrus may be quite low and the temperature gradient between cloud and ground not so marked then shadowing on the visual channels (bands 1 and 2) due to a low sun angle often highlights cloud edges and gives some indication of cloud top heights.

Middle and low–level cloud can often be difficult to discern on the imagery due to little or no thermal contrast between the cloud and ice surface. Under these circumstances the visual imagery (bands 1 and 2) again offer some clues through texture in the cloud tops from sun reflections and shadowing. With experience, surface ice texture, either in the sea ice zone, or over the continent becomes familiar as does textures associated with different cloud mass types.

On some occasions low cloud masses move in over the continent or sea ice zone with little or no discernible thermal contrast or notable texture. In these cases the cloud often may go completely undetected through analysis of the transmitted IR or visual radiometry data. However, during daylight hours under certain sun angle situations significant amounts of reflected solar long–wave radiation may be evident in the band 3 channel, caused by the energy at this wavelength being reflected by the large number of water droplets. Under these circumstances, areas of low cloud or fog, that are almost undetectable on the IR and visual channels, show up clearly in band 3 imagery as high reflected energy areas over the cold ice surface. For example, Figure 6.6.3.1.1 shows three bands (1, 3, and 4) from an AVHRR image taken at about 0927 UTC (around 1500–1600 local time) on 4 March 1997 over the Amery Ice Shelf–Prydz Bay area. It is obvious from the visible band (and to a slightly lesser extent from the infra–red band) that there is a band of cloud orientated northeast–southwest over the sea just left of centre of the image. It is not clear the extent to which this cloud band crosses the coast. The band–3 image, however, clearly shows that there is an extension of the cloud over the continent. Note that the cloud appears warmer (darker in this example due to the choice of grey–scale display) than the sea in this image because of the extra energy reaching the satellite sensor from the reflected light.

6.6.3.2                          Cloud Forecasting

Once areas of cloud have been detected the forecaster still has the task of forecasting the movement and development of the cloud mass. Or in the case of cloud–free areas, forecasting whether cloud is likely to develop. To achieve this goal the forecaster needs knowledge of the air masses in the area and predicted movements and development of systems.

There are several key points that are particularly relevant to Antarctic cloud including the following:

·                         When cold dry air moves out off the continent and over the water, heat and moisture fluxes give rise to low–level cloud some distance off the coast. This process may take some time to be accomplished, or may occur quickly. So even though a cold dry air mass is being forecast, if there is any open water present it is worth considering the likelihood of cloud formation. Sea smoke building into low cloud streets can form quite quickly in the sea ice zone if the wind begins to open up leads or polynyas.

·                         Northeasterly flow ahead of fronts and lows is the prime means of advecting moisture into coastal and inland areas. These features are generally easily identifiable on satellite imagery and normally well forecast by global numerical models. Smaller mesoscale cyclones in the coastal area may generate low–level cloud and be harder to both detect and forecast. These systems typically have life cycles of less than 24 h and can cause significant forecast problems.

·                         Some coastal stations occasionally (around four to six days a year) develop fog or areas of very low cloud associated with a sea breeze circulation. For example, typically on a fine light wind, sunny day at Casey a northwesterly flow develops in the area and over the period of several days advects, during the afternoons, moisture into the Casey area. Radiative and conductive cooling of the moist air where it hits the steep escarpment onto the plateau leads to the formation of a narrow bank of low cloud inland from the station. During the evening and overnight, a weak southerly outflow then advects the low cloud back down the escarpment and over the station as fog that may then last through part or most of the morning until another week sea–breeze circulation begins.

·                         With a lot of the smaller scale features, long–term forecasting techniques really do not exist and nowcasting and short–term trend type forecasts are all that can be done. On the broader scale, numerical model guidance is proving to be quite useful. Most models output from their diagnostic packages a percentage figure of low, middle and high level cloud, but another technique that is proving useful is a vertically integrated model temperature product starting from the model surface temperature field and using a threshold humidity value for cloud/no cloud detection at each model level. The product that results is a model synthetic IR image assuming perfect black body radiation from model assumed cloud. Currently this product is being produced at the Hobart Regional Forecasting center using a variety of NWP data and has proven very useful in cloud forecasting. While only showing the humidity predictions, Figure 6.6.3.2.1 is an example of NWP humidity forecasts for the flight path shown in Figure 6.6.2.1. And Figure 6.6.3.2.2 is an example of a synthetic cloud forecast (unrelated to Figure 6.6.3.2.1).

     Figure 6.6.3.1.1     AVHRR images taken at about 0927 UTC 4 March 1997 over the

     Amery Ice Shelf–Prydz Bay area.

Figure 6.6.3.2.2     An example of a synthetic cloud forecast. The forecast is from the       Australian Antarctic Limited Area Prediction System +24–hour prognosis for 1200 UTC 8 May 2002.       (Courtesy of Neil Adams, Australian Bureau of Meteorology.)

6.6.4                                Visibility and fog  

As in lower latitudes, fog and low stratus in the Antarctic may cause extremely dangerous conditions for aviation. Figure 6.6.4.1 shows a Pilatus Porter landing just before of a low stratus/fog bank invades the Mt King camp in Enderby Land (see Section 7.6.2). Arguably there are at least seven major mechanisms that relate to the formation and/or dispersal of fog and low stratus in the Antarctic region, these being:

·                         fog generated by air moving from open water over an ice surface;

·                         similarly, fog generated in the vicinity of melt–pools;

·                         "steaming" in which fog or mist forms as very cold air flows over the sea;

·                         the relationship between stratus/fog and orography;

·                         persistent stratus layer generated by turbulent mixing;

·                         stratus and fog close to the Antarctic Convergence;

·                         fog and stratus in the warm sector of low–pressure systems.

Aspects of the mechanisms mentioned above will be discussed in the following sections.

6.6.4.1                          Fog generated by air moving from open water over an ice surface 

Most often this process will occur over sea ice or fast ice due to the close proximity of these surfaces to a moisture source. However, the Antarctic coastal margin might also be affected under suitable circumstances.

Seasonality

The process of fog being generated by air moving from open water over an ice surface is most likely to occur in the beginning of the summer season. At this stage the sea ice is still quite extensive. Of course this depends upon the characteristics of the previous winter. Sea ice distribution is also dependent upon the configuration of the coastline: bays are well–protected locations that favour the persistence of the sea ice.

However, wherever open water is adjacent to an ice surface this process is likely to occur irrespective of season. This type of fog may form within a time span of two hours and turn a sunny and clear day, rather unexpectedly, into poor and murky conditions.

Process

Before moving over the ice surface, the air mass is warm and moist in the lower–most levels. As the air moves from the open water to over the ice surface, cooling of the lowest atmospheric levels occurs leading to the formation of a surface inversion. The presence of a considerable amount of moisture implies that, due to the cooling, condensation will take place quite rapidly.

Fog clearance

As long as the surface airflow steers the air from the sea to over the ice, this fog type is maintained. On average the visibility ranges between 200 and 600 m whilst the depth varies between 2 and 370 m (~8 and 1,200 ft). As soon as the wind changes direction so that the air is no longer blowing from the moisture source, the fog starts to disperse quite rapidly.

Locations where this process happens frequently

The above process probably happens on some occasions around the entire Antarctic marginal sea ice zone and adjacent coast. Areas of frequent occurrence are known to be:

·                         the Marguerite Bay–Rothera area (Section 7.3.7);

·                         the northern part of the Larsen Ice Shelf (Section 7.3.10);

·                         near the polynya between the Weddell Sea and the Ronne Ice Shelf (Section 7.4);

·                         the Brunt Ice Shelf–Halley–area (Section 7.5.2).

Forecasting considerations

Taking into account the information above, the forecaster will be mindful of the climatological predisposition for this type of process to occur in a particular area along with an assessment of the wind direction over suitable areas of adjacent open water and ice surfaces.

6.6.4.2                          Fog generated by melt–pools

This type of process is similar to the above to the extent that relative sources of moisture and coldness are adjacent to each other, however, in this case it is more a matter of air moving from a cold, dry surface over a warmer more moist one.

Seasonality

Melt–pools are pools of liquid water on an icy surface. Usually these phenomena form during the warmer part of summer, when, for example, in the Antarctic Peninsula area, positive temperatures are observed as far south as 72–73º S.

Process

The temperature of the melt–pool surface is higher than the adjacent ice surface. This temperature difference is variable but the larger the difference the more intense the following process will be:

·                         air moves from the ice to over the relative warmer melt–pool surface;

·                         as the temperature immediately above the melt–pool is higher than the adjacent ice surface evaporation from the top of the melt–pool starts;

·                         while over the melt–pool small scale convection occurs in the air mass;

·                         the resulting moisture in the air that has blown over the melt–pool is transported to the top of the surface inversion, eventually forming stratus as saturation is reached;

·                         as long as the temperature of the melt–pool surface is higher than the surrounding ice temperature, evaporation continues and the stratus spreads downwards, eventually reaching the surface to create fog;

·                         the depth of this saturated layer depends upon the height of the inversion generated by the radiational cooling of the lowest air layers over the general area. (There are likely to be subtler inversions set up when the air flows from the melt–pool area to back over the ice surface.)

Clearance of "melt–pool stratus/fog"

This feature can be very persistent, at least as long as a weak, variable wind prevails. Only a significant change within the synoptic situation is likely to lead to its clearance.

Locations where this event happens frequently

During the months of January, February and March many melt–pools are formed on the King George VI Sound ice shelf that is between Alexander Island and the southern part of the Antarctic Peninsula. As the British Antarctic Survey base Fossil Bluff (Section 7.3.8) is located in this area, BAS operations are frequently affected by this fog and stratus.

Forecasting considerations

Taking into account the information above the forecaster will be particularly mindful of synoptic situations that lead to clear skies and light winds over melt–pool regions.

6.6.4.3                          "Steaming fog"

Mainly affecting marine operations, and probably not often, is the occurrence where very cold air moves quickly over the relatively warm sea. Moisture evaporates into the air, which being very cold, soon reaches saturation and fog or mist forms. This type of event may occur when southerlies sweep off the Antarctic continent over polynyas, or where a ship might force open a lead in the sea ice. 

6.6.4.4                          Relationship between stratus/fog and orography

A smoothly ascending slope, typically towards a ridge or a dome, over which sufficiently moist air might flow, is a prime location for stratus/fog to occur. Typically a large area of stratus ascends gently towards a ridge or a dome. When the top is reached and the downward motion starts, the stratus begins to disperse. This is clearly shown by satellite imagery that reveals the lee edge of the stratus quite distinctively. As long as the surface flow direction is steady the lee side will remain stratus free.

Process

Cloud–free but moist air cools dry adiabatically (0.98ºC per 100 m) as it ascends the orography. Condensation occurs when the dew–point temperature is reached. The resulting stratus sheet is trapped beneath the surface inversion. Rising air continues to cool at the saturated adiabatic lapse rate, which is 0.6ºC per 100 m, due to the release of latent heat just above the cloud base.

The often sharp edge of the lee–side clearance of stratus may have two components to the cloud evaporation. The descending air warms at the dry adiabatic lapse rate, however, at the top of the ridge the potential temperature of the air has increased compared to the value the air mass had prior to impacting on the rising orography. This increase might occur in two ways. Firstly, should precipitation occur on the windward slope then the release of latent heat through condensation would warm the air (Foehn effect). Secondly, adiabatic compression may occur by the blocking of the air at the levels below the surface inversion on the windward side.

Locations where this process takes place regularly

The slope from Alexander Island towards the ridge to the north of Sky Hi (a fuel depot of BAS in the vicinity of the Sweeney Mountains–see Section 7.3.9 below ) is a favourite location as is the region around Siple Dome. The slopes of Berkner Island, on the Ronne Ice Shelf (see Section 7.4.3) also experience stratus/fog.

The example shown in Figure 6.6.4.1 is typical of a relatively common occurrence during summer of westerly airflow from the distant sea causing fog and low cloud at Mount King (Section 7.6.2).

Forecasting considerations

The key parameters to consider appear to be the presence of a moist airflow ascending a slope.  Knowledge of the upstream temperature and moisture may assist in determining the height at which the cloud might form.

6.6.4.5                          Persistent stratus layer generated by turbulent mixing

A relatively thin layer of non–precipitating low cloud (base often varies between 250 and 400 m (~1,000 and 1,400 ft), may persist for two to three days without any noticeable change.

Process

An explanation for this cloud can be found in the vertical wind distribution: a directional shear, varying between 100 and 180º, between the surface wind and the wind at the level above the cloud is the main reason. On top of this, considerable moisture content at this boundary level, should be present as well. At this boundary (level of directional shear) turbulent mixing results in constant moisture content throughout the depth of the mixed layer. This generates a saturated layer between the mixing condensation level (base) and the top of the mixed layer (top).

Forecasting considerations

As long as the directional shear exists turbulent mixing is present. Consequently the stratus layer persists. Indications of a change in wind direction might point to the clearance of this cloud layer.

6.6.4.6                          Fog and stratus in the vicinity of the Antarctic Convergence and in the warm sector of lows

The Antarctic Convergence is the boundary between the cold Antarctic waters to the south and the warmer sub–Antarctic waters to the north. Figure 6.6.4.6.1 shows the horizontal temperature gradient across the Antarctic Convergence while Figure 6.6.4.6.2 shows the position of the Antarctic Convergence (or equivalently the Antarctic Polar Front) around the Antarctic Continent. Note, for example the position of the northern parts of the Antarctic Peninsula with regard to the Antarctic Convergence, constrained in location of course by South America and the Peninsula itself.

Formation of fog and/or low stratus in the vicinity of the Antarctic Convergence.

This thermo–dynamic process is quite similar to the one of the water surface/ice surface case mentioned above. Air moves from north to south over the Southern Ocean and is continuously cooled in the lower–most levels. Consequently a persistent stratus layer is gradually generated. Across the Antarctic Convergence, which is characterised by a strong thermal gradient (Figure 6.6.4.6.1), the formation of fog is most likely.

Locations at which this event often affects operational activities

As the north of the Antarctic Peninsula is relatively close to the Antarctic Convergence, fog and low stratus affect frequently the weather conditions over there. Thus the impact of this phenomenon is quite significant as a lot of operational bases are positioned in that region. For example, Marsh and Marambio, the two alternate runways for the BAS's Dash–7 aircraft are positioned in that particular location (see Sections 7.3.2 and 7.3.5) Whenever the surface wind varies from westerly to northerly directions Marsh is likely to get fog. Marambio, positioned on the eastern side of the Peninsula gets fog or low stratus with a northerly to easterly surface wind.

Warm frontal fog/low stratus

The warm sector of lows, where the warm air is moving southwards over cooler seas, is a common location for fog/low stratus.

6.6.4.7                          Summary of forecasting considerations

The most important process to generate fog in the Antarctic is the advection of an air mass from a relatively warm surface onto a colder surface. This can happen within a space of time of about two hours that implies the importance of the detection of the fog in its incipient stage.

AVHRR satellite imagery is the most important information one can readily use to detect these fog patches quite rapidly. In order to obtain the most precise information both channels 2 and 3 should be checked. Channel 2 (reflected solar radiation) is less effective over a snow/ice surface. However, in channel 3 the snow/ice surface appears black so the white stratus or fog is very distinctive. To make a precise analysis of the satellite pictures a thorough knowledge of the orography is indispensable. And one should be aware of the sea ice distribution and its changes as well.

6.6.5                                Surface contrast  

The surface contrast is the ease with which features on a snow–covered surface can be distinguished, either from the air or by a surface observer. It is a very important element for travelling field parties who need to be able to make clear judgments about the state of the surface so as to avoid crevassed areas and large sastrugi fields. For aviation, a knowledge of surface contrast is also vital so that safe take offs and landings can be made.

Surface contrast is usually reported as good, moderate, poor or nil, although there are no internationally agreed definitions for each of these categories. The definitions used by the BAS are:

·                         Good: Surface features clearly defined;

·                         Moderate: Surface features visible, but become indistinct at more than a few kilometers;

·                         Poor: Surface features become indistinct at more than 50 m away;

·                         Nil: Footsteps and undulations discernible at no more than a few paces, if at all.

The surface contrast is dictated primarily by the cloud cover. In cloud–free conditions the surface contrast is usually excellent because of the small amounts of aerosol that are present in the Antarctic atmosphere and the very good visibility. When cirrus cloud is present the contrast can also be quite good. However, the greatest problems are encountered with deep, opaque layers of cloud, often in the form of featureless stratus, altostratus or nimbostratus. Under such conditions it is often not possible for a surface observer to see small mounds or crevasses on the surface when only a few metres away. When cloud with some convective elements is present, such as altocumulus or stratocumulus, the surface contrast is usually moderate or good. On the high Antarctic plateau cirrostratus can be quite near the surface and thick layers of this type of cloud can seriously reduce surface contrast.

The forecasting of surface contrast is therefore essentially one of forecasting the type and depth of cloud to be expected at a location. This subject is dealt with in depth in Section 6.6.3, however, here we will consider the aspects of cloud forecasting that are relevant to surface contrast. Forecasts of contrast are only usually required for up to about 12 hours ahead so a nowcasting approach can be used with satellite imagery playing an important role. Forecasts are made on a subjective basis, with the forecaster estimating the contrast based on his assessment of the cloud type and thickness. Determining the thickness of cloud is not easy, but the visible satellite imagery can be useful for estimating cloud thickness, since some surface features can often be seen through the cloud when it is thin. Short movie loops of visible images are also useful in forecasting cloud thickness since surface features can be seen more easily against the moving cloud field.

6.6.6                                Horizontal definition  

The horizontal definition is the ease with which the boundary between the ground and the sky can be determined. It is a parameter most appropriate over ice shelves or areas where there are no mountains or nunataks visible, which provide visual references on the horizon. The parameter is extremely important for flying operations on the ice shelves and in featureless parts of the continent, such as over the interior plateau. As with surface contrast, horizontal definition is observed to be either good, moderate, poor or nil and at the British Antarctic Survey these are taken to be:

·                         Good: Distinct horizon with obvious difference between land and sky;

·                         Moderate: Horizon visible but with no distinct difference in appearance of land and sky;

·                         Poor: Sky can be discriminated from land but no distinct horizon visible;

·                         Nil: Sky and land appear as one, no horizon visible.

In a similar way to surface contrast, the horizontal definition is determined primarily by the type and nature of the cloud present. Horizontal definition is excellent in cloud–free conditions, but gradually reduces as the amount of featureless stratiform cloud increases. The worst conditions are experienced when there is a thick layer of stratus, altostratus or nimbostratus. Horizontal definition is reasonable when the cloud has some convective elements, such as with altocumulus or stratocumulus. At the edge of an ice shelf the presence of an ice–free lead may aid the detection of the horizon. On the high plateau cirrostratus can be found close to the surface, which may reduce horizontal definition.

As with surface contrast, the forecasting of horizontal definition is carried out once a forecast has been made of the cloud to be expected at a location. The key elements required are predictions of the type and depth of cloud, plus knowledge of local orography, and these allow a subjective forecast of the horizontal definition to be made. As discussed under surface contrast, satellite imagery can be used successfully to estimate the type and thickness of cloud under many conditions.

6.6.7                                Precipitation  

The prediction of precipitation is very important for aviation and field party activities since it can reduce visibility considerably and make surface travel or work very dangerous. Moderate or heavy precipitation is almost always found in the coastal region of the Antarctic, while in the interior precipitation is usually light in intensity. These two areas also have fundamentally different types of precipitation. Near the coast most precipitation comes from frontal depressions, and to a lesser degree, mesoscale lows. In the interior most precipitation falls as clear sky precipitation or ‘diamond dust’, with few major synoptic scale lows reaching the area.

Near the coast the forecasts of precipitation from NWP models can be used directly to produce predictions of snow/rain amounts for several days ahead. Although we know that the current generation of NWP models have generally good representations of the large, synoptic–scale lows over the ocean areas, the precipitation forecasts should still be used with great care since the fine–scale structure, such as the locations of frontal bands, are not as good as those for mid–latitude areas. Whenever possible the model forecasts should be compared to the most recent satellite images to correct location errors of the frontal bands.

If only the model surface pressure forecasts are available then it is still possible to make some estimation of the locations of fronts by noting the sharp troughs in the surface pressure values around low–pressure systems. It is then possible to make a prediction of the likely arrival time of a frontal band, although the degree to which the front is active and therefore the amount of precipitation to be expected has to be estimated from the thickness field.

For short range prediction of precipitation in the coastal region (up to 12 hours, or exceptionally 24 hours, ahead) the most useful tool is visible and infra–red satellite imagery. Using sequences of images it is possible to forecast the arrival time of fronts at particular locations and to make some estimation of the likely precipitation from the structure of the frontal cloud band and the cloud top temperatures. Determining whether non–frontal cloud is precipitating or not from satellite imagery alone is much more difficult, although the cloud top temperatures can provide information on the cloud depth. For convective cloud, which is usually found over the ocean, the atmospheric temperature profile from a model or nearby radiosonde ascent is useful for estimating the likelihood of convective precipitation.

To determine whether precipitation will be in the form of rain or snow the model 500‑1000–hPa thickness values are the most useful parameter. At sea level, precipitation is usually in the form of snow (rain) when the 500–1000–hPa thickness is less (greater) than 528 dm, although consideration should also be taken of the detailed characteristics of the atmospheric temperature profile from a radiosonde ascent, if available. For example, when relatively warm air masses move over the Antarctic continent thickness values may be greater than 528 dm, but precipitation may still fall as snow because of the cold air temperatures near the surface.

Over the interior of the Antarctic prediction of precipitation is more difficult because of the lack of conventional weather systems. Diamond dust falls on many days, but because it is of such low intensity it is not a significant problem for operations. Model precipitation forecasts do not have any meaning in the interior of the Antarctic, but satellite imagery can be useful for following fronts as they penetrate inland, but any precipitation from these is usually slight once they are away from the coastal zone.

6.6.8                                Temperature and wind–chill factor  

Ironically, temperatures around the Antarctic are probably the least important parameter to forecast as all personnel will carry appropriate equipment and clothing as befits working in the Antarctic environment. Temperature forecasting in this area is performed mainly subjectively based on air–mass type and knowledge of sea temperatures and ice cover. Local snow melt during the summer revealing the rocks beneath, even over a limited area, can result in a surprisingly large diurnal variation under light wind conditions; the order of 4 to 5ºC is not unusual at Rothera for instance. An assessment of the temperature at remote sites under clear sky conditions can be obtained using satellite skin temperature data. This is graphically illustrated in Figure 6.6.8.1 where one senses the dome–like nature of the orography of Law Dome that is located near 67º S, 113º E simply from the IR grey–scale representation.

Wind chill, which was first considered by an Antarctic explorer (Paul Siple), is of course a significant factor when dealing with temperature forecasting, however local winds can be very difficult to forecast, particularly in the interior. Nevertheless, wind effects must be borne in mind, particularly when forecasting temperatures for remote sites. The approach to forecasting wind chill varies from location to location: for example, the British use the Steadman formula (Steadman 1971; 1979; 1984; 1994) for the relevant Antarctic Peninsula stations, while wind chill is not forecast for the Australian stations. The reason for the variation probably lies in the difficulty of the task. The reader is referred to an excellent review paper on wind chill by Maarouf and Bitzos (2000) that was written as a background document for the Wind chill Workshop (April 3–7, 2000) – hosted on the Internet (http://windchill.ec.gc.ca/) by the Meteorological Service of Canada.

To quote from Maarouf and Bitzos (2000), "Although wind chill indices are a useful way of quantifying the various detrimental effects of chilling winds, many researchers have indicated that wind chill indices used in Canada, the USA and elsewhere are overstated and should be adjusted. All wind chill indices in use today or published in refereed journals are based on methods that did not involve any human experiment or experience, which is a major flaw. Recent advances in heat transfer theory as well as modern experimental facilities such as wind tunnels and controlled cold chambers offer ample support for more rigorous research to evaluate and revise, if necessary, the present wind chill indices."

Figure 6.6.8.1     An AVHRR IR image taken at about 0929 UTC 15 February 1992. The circular nature of Law Dome (near 67º S 113 º E ) is evident from the grey–scale contrast.

6.6.9                                Aircraft icing  

In Section 3.4.2.6 a brief overview of the meteorological conditions associated with aircraft icing in the Antarctic is given, along with the effects on aircraft. Bundgaard (1951, p. 793) gives a succinct description of the physical process of aircraft icing: "For icing on aircraft in flight, the subfreezing cloud containing water in the liquid phase must be supersaturated with respect to ice. Although supercooled cloud droplets occur in air sub–saturated with respect to (a plane surface of) water, sublimation starts only after the moist air attains saturation equilibrium with respect to ice. In a layer sub–saturated with respect to ice, the physically unstable, supercooled fog droplets, upon striking the leading edges of the aircraft, are induced mechanically to congeal as light rime ice that almost instantly evaporates into the air streaming very rapidly over the rime–forming surface of the aircraft."

Examination of the sections on forecasting icing at the various Antarctic station given in Chapter 7 below suggests that aircraft icing is indeed a significant risk related to the presence of supercooled water droplets. The following quote, for example, is from the relevant section on icing for Casey (Section 7.10.1.4) "Forecasting airframe icing in Antarctica is quite difficult as an assessment needs to be made of whether the clouds are fully glaciated or whether some supercooled liquid may still be present. Certainly helicopters operating in the Australian sector of the Antarctic have experienced icing on numerous occasions so the forecaster needs to be aware. As a matter of course if there is any cloud present where the temperatures are above –20ºC then icing is mentioned. Severe icing is considered a possibility in pre–frontal cloud near the coast where it is possible that the airflow has been strong enough to carry supercooled liquid to the Antarctic coast".

The reference here to pre–frontal cloud is supported by Bernstein et al. (1997) who, in a study of the relationship between aircraft icing and synoptic–scale weather conditions for North America found that the second highest threat of icing occurred 250–600 km ahead of active and stationary warm fronts. (Arctic air masses were found to be in the highest risk category). Although not strictly relevant to the Antarctic, Bernstein et al. (1997) is recommended as a good resource of information on aircraft icing. These authors note that "Modern in–flight icing forecasters at the (USA) Aviation Weather Center (AWC) prepare their forecasts with the help of model–based icing algorithms……." (Bernstein et al. (1997, p. 742). They go on to indicate that even with these modern tools the AWC forecasters still need to hand–analyse synoptic scale weather patterns to diagnose the physical causes of icing and to interpret the NWP model data.

As an example of the state of the NWP art in this area Tremblay et al. (1995) have developed a technique to forecast supercooled cloud that, when coupled with NWP output of cloud water content, apparently leads to improved forecasts of aircraft icing.

6.6.9.1                          The –8D method  

It is not clear in the Antarctic context that the above level of sophistication available to the AWC is currently available, but with increasing access to Antarctic–focussed NWP this situation will change. At Neuymeyer (see "Icing" in Section 7.5.4.4) the analysis and forecasting of this parameter is prepared by using the radiosonde data processed by special software and by use of the so called "–8D–curve" method. This approach is the combination of a technique recommended half a century ago (see Bundgaard, 1951, p. 793) but using modern processing systems. In a reference quoted by Bundgaard (1951) it has been shown that supersaturation with respect to ice occurs whenever the air temperature is warmer than minus eight times the dew–point depression (D). When plotted on an aerological diagram if the "–8D" curve is located to the right of the observed temperature then icing is possible. Whereas no supercooled droplets exist when the –8D curve is left of the temperature. In Figure 6.6.9.1.1, for example, the "–8–D" curve indicates that a zone of potential supersaturation with respect to ice exists between about 800 and 700 hPa. Whether or not icing occurs will depend on the actual existence of the supercooled water droplets.

It should be pointed out that methods such as the one just described tend to be empirically based and, as indicated by Tremblay et al. (1995, p. 2,112) "typically tend to overestimate the presence of icing regions. This causes a tendency for pilots to ignore icing forecasts…".

6.6.9.2                          Detection of supercooled water droplets using AVHRR imagery

Sections 4.3.3.2 and 6.6.3.1 describe how to detect cloud that has supercooled water droplets in its upper levels, using, in particular, AVHRR band 3. In this band the reflected light from the supercooled droplets adds to the infra–red energy reaching the sensor thus giving a higher (than IR emission alone) count value.

Figure 6.6.9.1.1     A temperature profile (solid white line) and dew–point temperature profile (dashed white line) together with the low level "–8D" curve (red line).

6.6.9.3                          Detection of supercooled water droplets using SSM/I

In Section 4.3.3.6 products derived from the Special Sensor Microwave/Imager onboard the DMSP satellites is discussed. One of these products is cloud liquid water (see, for example, Figure 4.3.3.6.3). Tremblay et al. (1996) compare several techniques for the detection of supercooled liquid water (including the technique described in Trembley et al. (1995)) using SSM/I data to assist in the comparison. At a given horizontal point Tremblay et al. (1996, p. 70) assumed that supercooled liquid water existed if the cloud–top temperature was less than 0ºC, SSM/I retrievals of liquid water path (or cloud liquid water) exceeded 0.3 kg m^–2 and the SSM/I radiances were assessed as not being significantly influenced by ice crystal scattering.

6.6.10                            Turbulence 

Atmospheric turbulence forecasting is primarily needed in connection with aircraft operations and so where units of speed and altitude are mentioned in this section the aviation–preferred knots (kt) and feet (ft) will be used. There are arguably at least five types of turbulence with which a forecaster has to deal. These are turbulence associated with: (a) convection/squalls; (b) "mechanical" (strong winds blowing over terrain, usually at low altitude); (c) rotor streaming and low‑level lee waves; (d) mountain waves (at mid to high altitudes); and (e) jet streams/curved flow. Cloud may be present with any of these turbulence categories, or the turbulence may be in clear air (CAT). Moreover, as Bob Lunnon (Head of Aviation Applications, UK Meteorological Office points out in a personal communication, "the distinction between wind shear and turbulence is aircraft dependent so a forecaster should consider a mechanism that causes wind shear as a possible mechanism that causes turbulence". Relevant sources of wind shear include: temperature inversions; (synoptic) fronts; katabatic flows (land breezes); sea breezes; and hydraulic jumps. Apart from the last phenomenon (hydraulic jumps) there is little in the way of direct evidence of shear–related turbulence in the Antarctic context and so turbulence associated with these wind shear phenomena are not discussed here. However, from a theoretical standpoint, turbulence should not be unexpected in these wind shear producing features.

Whilst most of the literature deals with observations, theory, and forecasting (turbulence indices) for mid to low latitudes, there is no doubt that turbulence is a major factor for Antarctic aviation. For example, Lied (1968) reports on the crash of a helicopter that was operating at the then Wilkes Station (near the present day Casey Station (Section 7.10.1) on 13 February 1960. Lied reports that while in gale to storm force low level wind flow the aircraft experienced severe turbulence and vertical displacements of around 2,500 ft per minute. Even today aircraft in Antarctica that operate near rocky out–crops in strong winds; or fly down–wind of mountain ranges; or are en route at high levels to and from lower latitudes, all face a risk of encountering turbulence.

Therefore, this section briefly examines aspects of turbulence that might assist the Antarctic weather forecaster. Most of the information and guidelines are based on non–Antarctic data/experience. Moreover, Tony Skomina, an Aviation Specialist Meteorologist and turbulence expert with the Australian Bureau of Meteorology’s National Meteorological Operations Centre cautions that, in terms of clear air turbulence " it is essentially a sub–grid–scale process (microscale) and up till now we are using synoptic based predictors to isolate possible slabs of the atmosphere in which turbulence is present" (personal communication).

6.6.10.1                      Categories of the strength of turbulence

It may be inferred from Meteorological Office (1993, p. 113) that moderate to severe turbulence are the benchmark categories for describing turbulence. The definitions in Table 6.6.10.1.1 are taken from that reference.

In a personal communication Bob Lunnon (UK Meteorological Office) gives the following advice: "The quoted definitions (in Table 6.6.10.1.1) of turbulent severity are set in terms of the effect of the turbulence on an aircraft. The effects of turbulence on an aircraft are highly dependent on characteristics of the aircraft and the way it is flown, the most influential factors being aircraft weight and airspeed. Thus, given those quoted definitions, what might be light or moderate turbulence in a C–130 could well be severe or extreme turbulence in a Twin Otter. Thus if a forecaster is using this terminology it is important that he/she has a knowledge of what aircraft he/she is forecasting for. Ideally the forecaster should try and set up dialogue with the pilots and establish terminology that is consistent in its use by the forecaster and the pilot. If a pilot describes turbulence as "I wouldn't ever fly in that again if I had any choice" then this can be treated as a working turbulence severity for that aircraft type. The forecaster can calibrate his/her method based on such a working turbulence severity. It should always be remembered that, ultimately, the pilot is responsible for the safety of the aircraft, but if a forecaster can say that the circumstances are similar to those in which pilot X, flying aircraft Y, said  ‘I wouldn't ever fly in that again if I had any choice’ then this can be considered to be the best type of advice that can be offered to a pilot."

Table 6.6.10.1.1     A categorisation of turbulence.

Category of turbulence	Effect in/on aircraft
Light	Effects are less than those for Moderate intensity
Moderate	Moderate changes in aircraft attitude and/or height, but the aircraft remains controllable at all times. Air–speed variations are usually small. Changes in accelerometer readings of 0.5 to 1.0 g occur at the aircraft's centre of gravity. Occupants feel strain against seat belts. There is difficulty in walking. Loose objects move about.
Severe	Abrupt changes in aircraft attitude and/or height: the aircraft may be out of control for short periods. Air–speed variations are usually large. Changes in accelerometer readings of greater than 1.0 g occur at the aircraft's centre of gravity. Occupants are forced violently against seat belts. There is difficulty in walking. Loose objects are tossed about.
Extreme	Effects are more pronounced than for Severe intensity

6.6.10.2                      Clear air turbulence: indices and threshold values of wind shear

Richardson number

Probably the most well known index related to atmospheric turbulence is the Richardson number (Ri) as it is related to both shear and stability. (Meteorological Office (1991, p250)). Ellrod and Knapp (1992, p. 150) express Ri simply as:

R_i = static stability(vertical wind shear) ^–2                                            Equation 6.6.10.2.1

Intuitively it may be inferred from the above expression for R_i that turbulence would be associated with low values of this non–dimensional number: that is to say, one would expect turbulence to be associated with low static stability or with high vertical wind shear. According to Meteorological Office (1993, p. 116) turbulence is likely if R_i is less than 0.5 and is certain if R_i is less than 0.15. However, as Ellrod and Knapp (1992, p150) point out, low values of R_i are found in both turbulent and smooth regions.

Kelvin–Helmholtz instability

As described by Ellrod and Knapp (1992, p. 150) Kelvin–Helmholtz instability (KHI) is the principle mechanism found to be responsible for clear air turbulence: these authors liken KHI to the breaking of an ocean wave and note that KHI exists when vertical wind shear within a stable layer exceeds a critical value. Rees (1987), Kaneto (1982) and Kobayashi (1982) describe incidences of KHI in the context of strong stable stratification of the Antarctic boundary layer. Theoretical and numerical modelling discussions on KHI have been undertaken in a more general sense by authors such as: Sykes and Lewellen (1982); Fritts et al.(1996); and a companion paper by Palmer et al.(1996). Peltier and Scinocca (1990) examine the role of KHI in the pulsation of downslope wind–storms.

Recent indices for clear air turbulence are given by Ellrod and Knapp (1992):

TI1 = VWS* DEF                                                                                             Equation 6.6.10.2.2a

TI2 = VWS*(DEF + CVG)                                                                              Equation 6.6.10.2.2b

where, TI1 and TI2 are turbulence indices (1 and 2); VWS is the vertical wind shear; DEF the deformation of the flow; and CVG, convergence of the flow. The reader is encouraged to refer to Ellrod and Knapp (1992, p. 151 to 154) for an excellent description and explanation of the derivation of these indices. According to these authors "Prior research has correlated turbulence frequency to each term of the (index) equation and the three variables comprising the index (TI2) were found to contribute significantly to the turbulence potential" (Ellrod and Knapp,1992, p, 163).

However, David Thomas, the Australian NMOC Supervisor cautions in a personal communication that: "NMOC uses indices that include deformation as well as sheer (for example Ellrod and Knapp's TI1) but these type of indices are very sensitive to which numerical model output is used. Thus, absolute values are quite meaningless on their own with out some historical knowledge of the performance of the index. Every time we change a model it takes some time to reassess the critical values". These remarks are supported by Bob Lunnon (UK Meteorological Office) who, in a personal communication advises: "what works well over the contiguous USA may not work well in a relatively poorly observed region of the world such as Antarctica. We (the UK Meteorological Office) find that American predictors perform poorly almost anywhere in the world except the USA. We find that the "Brown predictor" (Brown, 1973) works as well as anything else in poorly observed parts of the world. We know that out own model (UK Meteorological Office) is not particularly good at forecasting vertical wind shear, and we suspect this is the reason "Brown" works well".

It would seem prudent then for Antarctic forecasters to undertake their own evaluations of the relevance in the Antarctic context of any of the turbulence indices that may be available.

Similarly, critical values of wind shear may vary from place to place. According to Ellrod and Knapp (1992, p. 153) vertical wind shear values of 6 kt per 1,000 ft are considered to be the threshold for moderate or greater turbulence. This is consistent with Meteorological Office (1993, p. 117) that adds that the vertical wind shear lower–bound for severe clear air turbulence is around 9 kt per 1,000 ft. David Thomas (NMOC) also advises that it was Australian practice to use, among a range of other considerations, this last mentioned criterion, however, experience has shown that a vertical shear criterion of around 20 kt per thousand feet seem more appropriate for predicting severe turbulence in the Australian context. So, again, the Antarctic forecaster will need to tune any such criteria listed below to suit the Antarctic experience.

Clear air turbulence: jet streams and curved flow

CAT may occur with most of the mechanisms discussed in the sections below, however, CAT is also often associated with jet streams (for example, (Shapiro, 1981)) and non–linear airflow. The turbulence index work of Ellrod and Knapp (1992) is specifically designed to provide an objective technique for diagnosing or predicting the occurrence of CAT where wind shear, divergence and convergence are kinematic contributors.

The Meteorological Office (1993, p. 116–117) provides some useful synoptic indicators of CAT that the Antarctic forecaster might also find useful. And, as pointed out by Bob Lunnon (UK Meteorological Office–personal communication) Kelvin–Helmholtz instability will occur when the Richardson number is low and, in synoptic terms, the conditions under which this occurs are those described in the above reference. These are:

·                         CAT associated with jet streams is most probably found:

–on the cold side, near and below the core;

–on the warm side, above the core;

–near exits with marked curvature and diffluence;

–at a confluence or diffluence of two jet streams;

–near sharp upper troughs;

–around sharp ridges on the warm side of jets;

–where one jet undercuts another;

–where the tropopause fluctuates.

·                         CAT is also found away from jet streams, but in areas of curved flow such as:

–in areas of anticyclonic curvature, where the actual wind speed approaches a critical value of twice the geostrophic wind speed. (Knox (1997) is an interesting reference concerning CAT and strongly anticyclonic flows);

–within 150 nm or so of the axis of a sharp upper trough where the wind shift is over 90º;

–occasionally, across shear lines in cols where the wind direction reverses rapidly.

6.6.10.3                      Convective turbulence

Turbulence due to convection in the Antarctic will probably be generally confined to rocky outcrops during days of strong insolation and may, or may not, be visible as cumuliform cloud. According to the Australian Bureau of Meteorology (1981, p. 44), convective turbulence usually causes no serious problems for aircraft except in thunderstorms. Due to the generally high stability of the Antarctic atmosphere cumulonimbus clouds are rarely if ever seen, and if they did occur would be confined to the coastline and Southern Ocean. Prydz Bay in Eastern Antarctica is believed to be one area where cumulus congestus bordering on cumulonimbus cloud infrequently occur when cold continental air mixes in a convergent manner over the relatively warm ocean. (See, for example, the discussion on clouds in Section 7.8.4.4.)

Mammatus cloud is also often associated with convective instability in mid to low latitudes and may occur in the Antarctic down stream of orographic ridges/hills/domes. However, in this context the mammatus may be more indicative of KHI and rotoring/lee wave activity (see Section 6.6.10.5) than of convection. While not a very clear image, Figure 6 presented in Adams (1996) shows an almost chaotic sky with lenticular, rotor and mammatus cloud in a strong wind and gravity wave event at Casey.

6.6.10.4                      "Mechanical" type turbulence

"Mechanical" type turbulence occurs in the lowest few hundred feet of the atmosphere and results from the frictional effects that strong winds experience over terrain. According to Meteorological Office (1993, p. 114) a surface wind of about 15 to 35 kt will cause moderate turbulence over flat country and severe turbulence over hilly orography while winds in excess of 35 kt will cause the turbulence to be severe and extreme respectively.

The Australian Bureau of Meteorology (1981, p. 45) cautions that strong winds will produce down–currents and turbulence in the lee of mountains irrespective of whether mountain waves have formed. This references advises aircraft approaching mountains into a strong head wind to allow adequate clearance from the ground to avoid stronger winds, turbulence and down currents near and in the lee of the peak (See also Sections 6.6.10.5 and 6.6.10.6).. Peltier and Scinocca (1990) is also recommended for aspects of downslope wind–storms in the lee of mountains.)

The forecaster will therefore want to know the terrain features (smooth snow; sastrugi; rocky outcrops, etc) over which any low altitude flying might take place and will need to forecast the surface wind speed.

6.6.10.5                      Rotor streaming and low altitude lee waves

Conditions in the Antarctic are probably ideal on occasions for rotor streaming in which a series of rotors form downwind of hills, ridges or mountains in a strong wind flow normal to the ridgeline. According to Meteorological Office (1993, p. 115), and Förchtgott (1969, pp. 256–257) (see also (Reece, 1979)) severe turbulence due to rotor streaming can occur when:

·                         strong winds (20 to 25 kt) occur near the ground at ridge level;

·                         a sharp decrease in wind speed, accompanied by a large change in wind direction, occurs about one and a half to twice the height of the ridge;

·                         a stable air mass occurs above the well–mixed lowest layer.

For an Antarctic forecaster, knowledge of the vertical wind profile away from main stations presents a problem unless there are in–field wind profile measurements available. From the top panel in Figure 6.6.10.5.1 it may be inferred that severe turbulence due to rotor streaming is confined to an area relatively close to, and downstream of, the mountain/hill ridge line and spans altitudes between just below ridge height to around two ridge heights. Förchtgott (1969, p. 258) also notes "besides extreme gustiness, the rotor zone is marked by local pressure deviations of several hectopascals that can easily escape the attention of pilots owing to their full occupation with the extreme effort needed to maintain the aircraft under a reasonable degree of control".

The bottom panel in Figure 6.6.10.5.1 suggests that low altitude lee wave formation is different from rotor streaming in that the lee waves can extend much further downstream of the ridge line with the severe turbulence zones more widely separated in the vertical and horizontal. The typical vertical wind profile is also different with a peak in horizontal wind speed normal to the ridge occurring at about four times the height of the ridge.

6.6.10.6                      Middle to high altitude turbulence associated with mountain waves

According to Meteorological Office (1993, p 116), mountain waves are probably the major reason for turbulence in the stratosphere, and while that part of the atmosphere is well above the area of interest for the Antarctic forecaster this illustrates the vertical extent to which mountain–generated turbulence might reach.

Also according to Meteorological Office (1993, p. 115), the intensity of the turbulence is proportional to the strength of the vertical motion and inversely proportional to the wavelength. The most turbulence–prone areas are likely to be near the wave crests and troughs. Check lists have been prepared for forecasting the occurrence of atmospheric turbulence, including mountain wave turbulence. The reader is referred to Ellrod (1989) for a comprehensive decision tree approach to diagnosing or predicting clear air turbulence including mountain wave induced turbulence (see also Section 6.6.10.2). This checklist, which requires access to upper–air data and to satellite imagery (including the water–vapour channel), is also available on the World Wide Web at http://www.met.fsu.edu/Ugrads/dwunder/prointro.html. A simpler checklist, which has found some utility in the Australian context, is shown in Figure 6.6.10.6.1 (Australian Bureau of Meteorology, 1973).

6.6.11                            Sea ice  

6.6.11.1                      Long–term considerations: atmosphere–ice interaction

Persistence of long wave troughs or ridges in one general area for any length of time (upwards of a week) will have an effect on the concentration and northward extent of the sea ice through the long wave modulation of synoptic features. If a long wave ridge persists over the marginal ice zone the ice pack should become more consolidated due to the formation of new ice and refreezing of old ice. This is due to more frequent periods of calm or light winds, clearer skies and colder temperatures associated with short wave features in the ridge, than if a long wave trough were located in the area.

If, on the other hand, a long wave trough does persist in the area the ice pack should be more broken with heavy rafting of floes and large fields of brash ice due to the constant motion of floes under the action of the wind. If the area is on the western side of the long wave trough then there should be a tendency for the ice edge to be further north than on the eastern side of the trough due to the winds having a greater southerly component. The ice edge on the western side of the trough will be more fragmented due to the divergent nature of this motion than that on the eastern side where the winds will be more northerly and act to concentrate the floes.

Strong winds and large amplitude swell have a great bearing on the motion of the ice in the Antarctic pack. Predictions of persistence of periods of severe or calm weather can greatly assist mariners by giving guidance on the suitability of operating in any given area and the risk of becoming trapped there for extended periods of time.

6.6.11.2                      Short term considerations

Nowcasting

The sea ice analysis services (see Section 2.8.6) together with any available satellite imagery of sea ice conditions, may allow a very short-term forecast of ice conditions to be provided. However, the effects of wind and currents may limit the usefulness of these data, particularly very high–resolution data.

Forecasts

The NIC provides a seasonal Ross Sea (90‑day) forecast for the McMurdo Sound resupply: this forecast is available on 15 October of each year. (http://polar.ncep.noaa.gov/seaice/). The NCEP Ocean Modeling Branch provides input to the NIC products as well as producing sea ice forecasts themselves.

Vessel Icing

An algorithm is given by Overland et al., (1986, p. 1801) relating sub–freezing air temperatures, strong winds and sea surface temperatures of less than 6⁰C to superstructure icing on ships:

PR = V(T_f–T_a)[1+0.4(T_s–T_f)] ^–1                                                                       Equation 6.6.11.3

where:                         V is wind speed in ms^–1;

T_f is the freezing point of seawater (ºC);

T_a is the air temperature (ºC);

T_s is the sea surface temperature (ºC);

PR is a predictor that may be used to give ice accretion rates using Table 6.6.11.3.1.

Table 6.6.11.3.1     Categorical forecast ranges for vessel icing.

	Light	Moderate	Heavy
Predictor (PR) (m ºC s–1)	< 20.6	20.6 to 45.2	> 45.2
Accretion rate (cm.hr–1)	<0.7	0.7 to 2.0	>2.0

6.6.12                            Waves and swell  

Forecasts of sea state are required in support of the operation of vessels of all sizes, ranging from ships on the open ocean engaged in commercial and resupply activities to small boats used in areas of open water inside the sea ice zone. The nature of this publication dictates that the treatment be extremely brief. The reader is referred to WMO (1998) from which much of this material is derived, for greater detail.

It is normal to describe the state of the sea in terms of a combination of sinusoidal waveforms. Such waves are classified according to their period and range from long period waves, such as tides and tsunamis, to extremely short period capillary waves, dominated by the surface tension of the water. The discussion here will be confined to wind generated gravity waves with periods generally ranging from 1 to 30 seconds. The derivation of the relationships regarding wave motions is beyond the scope of this publication and can be found in any standard text on wave motion (for example Crapper, 1984). Before progressing to a consideration of the state of the sea in its entirety it is useful to consider the properties of a unidirectional periodic wave.

6.6.12.1                      Basic Definitions and Relationships

While it is beyond the scope of this handbook to explore the relationships between the following wave characteristic descriptors in detail it is considered useful to briefly list the key wave parameters:

·                         The wavelength (λ) is the horizontal distance between successive wave crests;

·                         The period (T) is the time between two successive crests passing a fixed point;

·                         The frequency (f) is the number of wave crests that pass a fixed point in unit time and is the inverse of the period;

·                         The amplitude (a) is the maximum displacement from the equilibrium position;

·                         Wave height (H) is the vertical distance between a wave trough and the following wave crest. In the case of a sinusoidal wave it is twice the amplitude;

·                         The phase speed (c) is the rate at which the wave form advances and is equal to λ/T;

·                         The energy propagates at the group velocity (C_g) that, in deep water, is half the phase speed.

6.6.12.2                      Effects of Water Depth

As waves move into shallow water only the period will remain constant while the phase speed and wavelength decrease. Water depth (H) is normally classified as deep, transitional or shallow depending on the ratio of depth to wavelength. In deep water the depth exceeds one quarter of the wavelength and in shallow water the depth is less than one twenty–fifth of the wavelength.

6.6.12.3                      Combinations of waves

As stated earlier, the total sea state can be described in terms of a linear combination of the simple waves described above with varying amplitude, frequency, phase and direction. These waves may be generated locally by energy input due to the surface wind stress (wind waves), or propagate into the area from afar (swell waves).

It is convenient to define another parameter, the significant wave weight, which is the average height of the highest third of all waves. This quantity is one of the outputs of numerical wave models and also usually corresponds to visual observations of the sea state. It must be strongly emphasised that waves larger than the significant wave height will be encountered and the maximum wave is likely to be approximately twice the significant wave height. Even larger waves are possible.

In this spectrum of waves the variance of the surface elevation caused by each component represents the contribution to the total energy by that component, as explained above, hence the wave spectrum also represents an energy spectrum. Various functional forms for this spectrum have been proposed such as the Pierson–Moskowitz (Pierson and Moskowitz, 1964) and JONSWAP (Hasselman et al., 1973).

A practical side-effect of the spectral nature of waves is that various vessel types respond differently to the wave period: as a result wave model output of the peak energy period (T_peak or T_p) may be more useful than the significant wave period.

6.6.12.4                      Effects of sea ice

The effects of sea ice are difficult to quantify, but even the very thinnest of newly formed ice progressively removes the higher frequency, small amplitude waves as it develops and thickens. Very high concentrations of mature ice can effectively damp out all but the longest period waves. Romanov (1996) reports that the distance of swell penetration into ice cover, evaluated from satellite observations, is from 300 to 400 km and that swell waves may reach the Antarctic coast even at the time of maximum ice development.

Large recurring polynyas commence forming around the Antarctic coast about September, reach their maximum extent around January and decrease to the September level by May (Romanov, 1996). It is the presence of these large areas of quasi–open water within the sea ice zone that leads to a natural dichotomy in methods of wave forecasting. In particular, some global wave models do not include pack ice and will thus not model dampened wave motion, as they should.

6.6.12.5                      Numerical wave models

The basic principle of numerical models is to calculate the changes with time of the energy spectrum mentioned above. Models containing the region of interest to Antarctic forecasters are run on the global domain at resolutions between one and three degrees and normally in deep water mode without currents.

Without going into the specifics the total rate of change of energy equation can be written such that the energy source terms are:

·                         the energy input by the surface wind stress;

·                         non–linear transfer due to wave to wave interactions;

·                         and dissipation due to whitecapping.

The initial state is normally a forecast from a previous model run and the wind forcing is provided by the output from a numerical atmospheric model. The major source of error in numerical wave forecasts lies in errors in the input wind fields (Komen and Smith, 1999).

Verification data over the Southern Ocean are scarce, but preliminary results obtained in NMOC, Melbourne comparing the output of the GASP model (Bourke et al., 1990) with observations from the vessel Aurora Australis (Table 6.6.12.5.1) indicate an acceptable level of correlation for forecasts of up to 60 hours, with a decline for longer intervals.

Although Romeiser (1993) reported an under–estimation of wave heights in the Southern Ocean during winter by the wave model–WAM (Hasselman et al., 1988) recent results obtained in NMOC Melbourne indicate a tendency for the Australian implementation of WAM to overestimate wave heights during strong wind events, resulting in a high bias over the Southern Indian Ocean.

The other major source of error in high latitudes is the presence of sea ice, the treatment of which varies between centres. The ECMWF implementation of WAM allows for a variable ice edge, based on analysed sea surface temperature that is then held fixed for the duration of the forecast, whereas the Australian implementation of WAM currently takes no account of sea ice at all.

Table 6.6.12.5.1     Correlation between

GASP forecast wind speed and observations

from Aurora Australis.

Forecast Interval (hours)	Correlation Coefficient
0	0.89
12	0.87
24	0.84
36	0.85
48	0.82
60	0.78
72	0.61
96	0.67

6.6.12.6                      Manual wave forecasting methods

Although manual wave forecasting methods can be applied over the open ocean, they are labour‑intensive and forecasts for these regions are more conveniently derived from the output of numerical models. However, their use is mandatory in large polynyas within the sea ice.

Manual techniques rely principally on the use of nomograms to determine wave growth without consideration of the physical processes involved. They have largely been devised empirically from observed or measured wave datasets and rely on pre–determined properties of the wind field (fetch, duration and wind speed and direction). It is therefore immediately obvious that the first steps are the estimation of the wind field and ice conditions. These processes are covered elsewhere in this handbook and will not be repeated.

Having determined the likely fetch, duration and wind field it is a simple matter to determine the wave height from nomograms: see, for example, US Army Coastal Engineering Research Center (1973). If a functional form of the wave energy spectrum, as discussed in Section 6.6.12.3, is assumed then information regarding the range of waves present can also be calculated.

As the wind field is unlikely to remain constant during the forecast period some empirical “rules of thumb” are useful in modifying the forecast wave field:

·                         (i): If the wind freshens at constant direction, subtract one quarter of the increase from the new wind speed and use this value to determine the wave height from the nomogram. For instance, if the wind speed is predicted to increase from 10 to 20 m s^–1 over a 12 hour period compute the wave height using a speed of 17.5 m s^–1 over a duration of 12 hours.

·                         (ii): Changes in wind direction at constant speed of less than 30º can be ignored and the direction treated as constant. For greater changes the previous wave field should be treated as swell and it becomes necessary to perform another calculation of the newly generated sea state.

·                         (iii): Once the wind slackens swell wave heights can be reduced by 25% per 12‑hour period.

6.6.13                            Hydraulic jumps (Loewe's phenomena)  

One of the most, if not the most, dramatic feature of the coastal climate of Antarctica is the very strong katabatic wind blowing frequently from the polar plateau toward the sea. Apart from the severe downslope wind itself, the most spectacular phenomenon occurring during the katabatic periods is the "Loewe's phenomenon". This type of phenomenon has been described by Valtat (1960) as occurring at Dumont d'Urville, Adélie Land. The Loewe's phenomenon includes a sudden slowing down of the wind speed and a change in the depth of the cold air layer. This "jump" in the wind is always associated with a "jump" in the pressure so that, sometimes, the Loewe's phenomenon is just called a "pressure jump". Drifting snow is also always observed upstream of the jump, as the wind is strong. A simple theory derived from the hydraulic channel flow theory has been developed by Ball (1956), from which various characteristics of the flow can be predicted, including the conditions at a jump occurring as a transition from shooting to tranquil flow. By analogy with the hydraulic theory, the Loewe's phenomenon is then sometime called "hydraulic jump" or jump.

The IAGO (Interaction–Atmosphère–Glace–Océan) experiment was aimed at obtaining a detailed documentation of the vertical structure of the katabatic layer, and of its evolution along the slope. Two French and one American teams took simultaneous measurements of vertical profiles of atmospheric properties at three points (D57, D47 and D10) distributed over 200 km along the slope from the plateau, at about 2,000 m height (~6,500 ft), toward the coast near the Dumont d'Urville (DDU) Station (see Figure 6.6.13.1).

During the IAGO experiment several Loewe's phenomena were observed as the terminal phase of several katabatic events. On 3 December 1985, a surprisingly large surface–pressure change (almost 6 hPa) through a jump was measured, differing very strongly from the predicted value (about 2 hPa) derived from the hydraulic theory. This case is not isolated, as Lied (1964) reported similar observations (Lied measured 20 hPa through a jump on 12 August 1961, near Davis Station (see Section 7.8.4).

Although quite often quoted in reports from Antarctica explorers, relatively few Loewe's phenomena or jumps occurring through katabatic flows have been scientifically described. Obviously, Antarctica is not an ordinary experimental ground and, apart from the intense cold, the observation conditions of such phenomena are extremely severe. Nevertheless, these jumps can be considered as a common feature occurring in these regions.

Nevertheless, a typical evolution of katabatic events and of Loewe's phenomena can be drawn from the literature descriptions. The gravity flow is strengthened if the pressure in the air above the inversion decreases horizontally in the downslope direction, while it is weaker and can be completely stopped if the pressure in the overlying air increases instead. The formation of a jump happens just before a strong katabatic event and after the katabatic flow has persisted for several hours or days and generally marks the beginning and the end of the katabatic period. The jump can be defined as a narrow zone with large horizontal changes in wind speed, pressure and temperature, either stationary on the slope or slowly moving up or down. This zone separates between an upstream shallow, strong–wind layer, where the flow is said to be "shooting", and a downstream, deep, light–wind layer, where the flow is said to be "tranquil". A wall of snow often very clearly marks the downstream edge of the strong wind part of the flow.

Figure 6.6.13.2 shows a schematic of the jump observed on December 3, 1985, at the D10 Station. A wind reversal was recorded just downstream of the wall of snow, which then moved upslope after having passed the station. This reversal is possibly associated with a convergence effect and an updraft developing through the jump, marked by the formation of clouds situated just above, and moving with, the jump.

Figure 6.6.13.2     Sketch of the Loewe phenomenon that occurred on December 3, 1985, at D10 near Dumont d’Urville. (F represents the Froude number.)

Figure 6.6.13.3     Surface measurements of pressure (solid circles), potential         temperature (open circles), and downslope wind speed (crosses) as a function of local time during the phenomenon.

Figure 6.6.13.3 shows the surface measurements as the jump moved slowly upslope. The southerly surface wind suddenly dropped from 20 m s^–1 to –5 m s^–1 as pressure simultaneously increased from 958.5 hPa to 964.2 hPa. Some minutes later, potential temperature increased from 264.6 K to 267.0 K, probably due to turbulent mixing with upper–air associated with the jump.

Figure 6.6.13.4 shows the vertical profiles of the downslope component of the wind, before and after the jump, also interpreted as upstream and downstream conditions. The flow is vertically stratified from the surface upwards with:

·                         a cold–air surface–layer, either well–mixed or with a small Richardson number;

·                         a very stable capping inversion layer;

·                         an unstable layer thickening as the flow goes downstream;

·                         a stable transition layer to the overlying free atmosphere.

The "katabatic layer" is composed of the well–mixed layer and of the capping inversion layer, where the maximum of wind speed is observed.

6.6.13.1                      Pressure change through Loewe's phenomena:

Even taking into account the perturbations associated with upstream blowing snow, the surface–pressure change mainly scales with the shear observed through the unstable layer overlying the katabatic layer. As always mentioned by people having observed Loewe's phenomena, snow transported by strong katabatic flows can indeed be suspected to be an important factor contributing to this problem, and it can be argued safely that the largest surface–pressure changes are observed with blowing–snow conditions. An interpretation is that Antarctic katabatic flows are a perturbed state of the atmosphere, in which surface pressure is reduced in comparison to what it would be for unperturbed conditions, which are restored just downstream of the jumps whenever they occur. This reduction of pressure is mainly associated with the strong acceleration of the katabatic layer under the effect of gravity.

These conclusions are consistent with pressure changes associated with other observed Antarctic jumps. Thirty–one cases of standing katabatic jumps were reported and measured by Lied (1964) during 1961 on the slopes of the polar ice cap in Antarctica. Quoting Lied's words: "In most cases the pressure changed by 1 to 3 hPa through the jump, but on one occasion a mysteriously large change of nearly 20 hPa was measured". In fact, that surprisingly large change can be associated with a strong wind shear of 60 m s^–1 (~115 kt) between the surface and the top of the well–mixed layer over the katabatic layer. Such a large pressure change is obviously exceptional, but still likely to happen considering the very strong winds occurring during katabatic events in Antarctica.

Such evidence for a layered structure of the upstream part of katabatic flows is an important feature for Antarctic katabatic flow modelling. Upstream of the jump, the vertical structure of the flow is relatively simple and exhibit, a four–layer structure, i.e. katabatic layer, inversion layer, overlying neutral or unstable layer, and above free flow. This makes it possible to use simplifying assumption, of the type used by Ball, and explains the success of Ball's hydraulic theory for describing strong katabatic flows. On the contrary, observations show that Loewe's phenomena have a frontal character and that downstream of them the flow loses its simple behaviour in such a manner that Ball's assumptions cannot be used any more, explaining then the failure of the hydraulic theory.