Issues Magazine

Forecasting Antarctic Weather

By Scott Carpentier

A weather forecast is never a guarantee, but activities in the unforgiving Antarctic environment become much more risky without one.

Humans are poorly designed to survive the cold and harsh Antarctic environment. Many Antarctic tragedies occur through exposure to the elements, specifically low temperatures and high winds or, worse still, the combination of the two known as the wind chill factor. With high winds can come the associated loss of visibility and potential disorientation when snow is lifted above eye level during a blizzard.

The waters of the Southern Ocean near the Antarctic coast can reach temperatures near –2ºC. Falling into such cold water can lead to cardiac arrest, or slow neuromuscular activity which can quickly lead to loss of coordination and ultimately drowning. Typically, this can occur within 10 minutes if the person is not wearing an immersion suit.

On land, the major risks are associated with hypothermia: the slow loss of core body temperature that leads to the failure of major organs and brain death. In Antarctica, hypothermia typically occurs to expeditioners losing their way in blizzard conditions without appropriate means to take shelter.

In his book about the first Australasian Antarctic Expedition (1911–1914), Home of the Blizzard, expedition leader Douglas Mawson describes the harshness of the environment and what is now known as the wind chill factor:

Temperatures as low as –28 degrees F. [–60ºC] were experienced in hurricane winds, which blew at a velocity occasionally exceeding one hundred miles per hour [160 km/h]. Still air and low temperatures, or high winds and moderate temperatures, are well enough; but the combination of high winds and low temperatures is difficult to bear.

In the modern era, outside of the cosy confines of our well-insulated Antarctic stations, the three elements of cold, wind and poor visibility continue to have an impact on operations. For example, in the Australian Antarctic Program (AAP) during December 2011 alone:

• two expeditioners became disorientated and lost for some hours within metres of their field camp during a blizzard. By chance, the blizzard briefly abated, allowing them to find their way back to camp. They suffered the beginnings of hypothermia and extensive frostbite to their faces, hands and feet;

• two fixed-wing aircraft bound for Dumont d’Urville waited in Hobart for 4 weeks before a favourable window of weather presented itself; and

• the planned 9-day ship-to-shore resupply of Casey Station (refuelling, personnel deployment/retrieval etc.) took 20 days as bad weather, including 6 days of blizzard conditions, affected the operations.

Antarctic weather forecasts minimise the likelihood of injury to expeditioners and loss of time due to bad weather. The challenge of the operational forecaster is to provide timely and accurate forecasts that are of service to the users, be they aboard a ship in fog-prone ice-covered Antarctic waters, in a small aircraft operating at low altitudes along the coast, in a crane swinging shipping containers on a wharf, in a ski party on their way to a field hut, counting penguins on sea-ice or simply working on a roof. Without the knowledge of a weather forecast, all those activities become more risky.

Observation and Analysis

The foundation of a good weather forecast is the observation network: it is only possible to tell what will happen tomorrow by first knowing what is happening now. To answer the question of what is happening now, there is a coordinated global observation network. It consists of a surface-based array of automatic weather stations (AWS), human observers, radars and wind profilers; and an airborne program of radiosondes (weather balloons), weather instruments attached to airplanes, and satellite-derived observations.

In each and all of these components, the Antarctic network is the most sparse and deficient of anywhere on the planet. For example, a coordinated global release of radiosondes at 23 UTC (coordinated universal time) and 11 UTC takes high-resolution measurements of wind, temperature and humidity. The sondes routinely provide transects of the atmosphere up to a height of 30 km. There are 26 upper air radiosonde stations south of 45S, compared to 370 stations over an equivalent region in the Northern Hemisphere. A three-dimensional analysis of the atmosphere based on these transects is clearly superior in the Northern Hemisphere, while analysis of the high southern latitudes is more error-prone.

Radiosondes are also difficult to release in high winds such as those that frequently occur at Antarctic coastal stations, so many extreme events are not captured by sonde releases. On the very morning that I write this, the Casey Station observer relates:

Hi all: NOREP [no report] sent for this morning’s flight at Casey.

First sonde – temperature sensor failed GC [ground check].

Second sonde – delayed launch as needed to dig snow from inside and outside balloon shed in order to open doors. String mast snapped and sonde hit ground on release in about 65G73KT winds [120 km/h gusting to 135 km/h]. More snow digging required to close balloon shed doors again after failed launch.

The second sonde may still be good to use another day once we replace the string mast, but no further attempts will be made at releases this morning due to the high likelihood of another failed launch.


TO2 Observer, Casey Station

The environment is particularly harsh on the AWS network, and repairs in the more remote deep field locations may take months if not years, making the already sparse observation network even patchier and more error prone. Typical AWS malfunctions include broken wind cups and wind vanes due to high winds, the AWS being buried in snow or rimed up in ice, faulty electronics and batteries due to low temperatures.

Numerical Weather Prediction

Numerical weather prediction (NWP) models treat the atmosphere as a fluid whose motion is governed by three main sets of equations: conservation of mass, conservation of momentum and a thermal energy equation. These are collectively known as the primitive equations.

The first task of NWP models is to define an initial state of the atmosphere (called initialisation), based on the observational snapshot already described. The next task is to compute the future state of the fluid by solving the primitive equations. This is not trivial, because these equations are impossible to solve directly through analytical methods. They are therefore supplemented by simplifications to account for sub-grid processes or processes that are too complex to be explicitly included in the model.

The main limitations of NWP in simulating Antarctic weather include:

• model initialisation, which is based on a sparse and error-prone observation network and thus more likely to develop chaotic responses sooner;

• air–sea exchanges of heat and moisture over sea-ice, which are poorly represented due to both coarse resolution and lack of ice thickness information. This is particularly a problem in winter, when up to 18 million km2 of ocean is covered by sea-ice (1.7 times the size of the Australian continent); and

• cloud microphysics schemes, which poorly represent clear air precipitation (diamond dust). Diamond dust is ice crystals that form like fog, but only in very cold temperatures. It accounts for a significant proportion of precipitation on the continent.

Thus the Antarctic weather forecaster is required to have a sound working knowledge of NWP models, from the quality of the observations it ingests into its analysis scheme, the limitations imposed on it through its resolution, and the assumptions of its simplified equation schemes.

Satellite Imagery

There are two types of weather satellites: geostationary and polar-orbiting. Geostationary satellites are positioned over the Equator, so the Antarctic tends to be towards the limit of their field-of-view, resulting in a “fuzzy” side-on glance at the horizon rather than a more optimal birds-eye view of the sky. Although geostationary satellites do not resolve Antarctic cloud fields particularly well, they are still useful for monitoring large-scale (synoptic) features by providing hourly images over the Southern Ocean and as far south as the Antarctic coast.

Because they fly overhead, polar-orbiting satellites provide far better spatial resolution of Antarctic cloud fields. The biggest issue that affects the use of polar-orbiting data is their small footprint and long periods between over-flights. It is not unusual to have to wait 2–3 hours for the next useful polar-orbital pass. Discerning white clouds from the white icy background can be another source of frustration for a forecaster.

Weather satellite images and their interpretation is a mainstay of the forecast process. From these, cloud fields can also be compared against model simulations, thereby showing how well the NWP model is representing the present conditions.

Antarctic clouds are particularly poorly modelled due to the poor observation network, the cold temperatures, the expansive regions of suspended ice crystals blown up from the wind that may not have been resolved, and possibly also due to the distinctly unpolluted air.

Conveying Uncertainty

The future is never certain. We have expectations, largely based on prior experiences, but no one can tell what will happen next week, tomorrow, or even in a couple of minutes with complete confidence. If a pilot in an aircraft encounters weather conditions that are worse than anticipated, like stronger headwinds, a decision needs to be quickly made on whether to continue as planned and take the risk of running out of fuel, or turn back. If the forecast at pre-departure was a categorical statement of headwinds of 15 km/h, an unwise pilot may fuel up to a level that will get them to their destination based on conditions slowing them down by 15 km/h. The forecast could, however, convey some uncertainty by stating headwinds of 10–20 km/h. The pilot would then consider conditions of 20 km/h headwinds as a possibility and fuel up accordingly. Pilots are, of course, more conservative than this and by law need to carry extra fuel reserves in the eventuality that conditions don’t pan out as anticipated.

Educating our clients on critical thresholds of their local weather is also important. If the forecast is for 40–50 km/h wind speeds, a well-briefed trip leader of a land traverse would plan a contingency based on winds being a little stronger than forecast. This wise trip leader would then recognise that should the winds rise to 60 km/h, reductions in visibility may also develop due to snow being lifted off the icy surface. Such conditions could hamper travel and possibly even lead to disorientation of the group.

Low-speed Communication Links

The volume of data required to support a modern forecast office is quite high, and presents a major challenge for the computing managers responsible for data transfer to Antarctic stations. Satellite links are expensive and slower than the broadband networks available in most modern countries nowadays. For example, the Bureau of Meteorology’s Hobart office datalink works at a nominal speed of 10 Mb/second, whereas the Davis Station office runs 100 times slower! Therefore, Antarctic forecasting offices have to make do with slower computer visualisations and cut-down versions of NWP models. This can be a major source of frustration when deadlines are looming or complex situations are poorly resolved.

Future Challenges

Funding for weather research is largely driven by public and industrial agencies concerned with impacts to their operations, property and life. For example, a 150 km/h windstorm along the east Australian coast would likely have a far greater societal impact than a similar storm on the Antarctic coast, so it is no surprise that research is focused on tropical and mid-latitude meteorology.

When it comes to research, Antarctica is still the great frontier. Challenges for the future include:

• improving the observation network;

• improving sea-ice analysis and forecasting;

• better resolving the interactions at the boundary between moist-maritime and dry-continental air masses;

• improving modelling of clear air precipitation and cloud microphysics;

• improving understanding of how Antarctic weather systems interact with systems further afield; and

• continuing research on the evolution of the Antarctic climate in a global warming world.

Notwithstanding all the challenges so far, forecasting for the Antarctic is a highly rewarding experience. The relationship between the meteorological community and the international polar effort is longstanding and based on good will. It is also a relationship built out of necessity, and ultimately based on safety to life and property.

The Antarctic community is relatively small and specialised, so polar forecasters tend to know their clients personally. There is an underlying sense of collaboration with every Antarctic expedition that we support, encapsulated in this message from this year’s station leader at Mawson:

Hello Scott,

On behalf of all of us at Mawson I am very grateful for the weather forecasts that you provided. After nearly constant gale force winds in June, we were able to take advantage of 4 fine days for the return trip and to conduct the photography at Taylor Glacier Emperor Penguin Colony on Monday 2nd July.

Thank you again for assisting us in being able to conduct this work safely.

Best wishes, Bob.

Station Leader MAWSON, Antarctica