Time is always on the move, but every day in Australia, it meets at 9.45 am.
Having emitted high temperatures for each of their states in the early hours of the morning, meteorologists from all over the country connect to the national headquarters of the Bureau of Meteorology to ensure that each office starts the day by reading the same script.
In HQ, the eyes of about 50 forecasters are glued to the multi-screen display in the front of the room as the head forecast for the day begins, indicating the pressure systems in motion and a large cold front near Perth. It is May and even Western Australia is not immune to rain.
With the increase in computer power and observational data, we can now predict seven days earlier with accuracy similar to three days at the beginning of the century – but how?
We know that technology is essential but what kind exactly? Does Melbourne really have "four seasons in a day"? How does the familiarity of a forecaster with the summer blues south of Sydney come into play? In the era of supercomputers and satellites, how far can we predict our time?
How do meteorologists keep up with time?
At the national conference of the Bureau of Meterology, the extreme weather desk is next to the report. They show seven orange lines off the northern coast of Queensland – each a computer generated prediction of the path of a tropical cyclone – moving away from the coast. "All lines follow a similar path, so we can be sure where it is going," announces the forecaster.
Meteorologists working on aviation and oceans follow. The last is the desk of the volcano, which monitors the circulation of volcanic ash in the world. The Indonesian volcano Dukono is spewing small amounts, as usual.
Hidden in a radio studio, another meteorologist is chatting on the ABC radio. Many forecasters listen carefully to the conference, at least in part because they want to avoid being caught by an unexpected question during their television or radio appearances later that day.
Getting the right forecasts matters. Not that the BoM always does: a few years ago, they predicted an El Nino system and its dry climate that accompanied it for central NSW, according to NSW farmer Wayne Dunford.
"But that year, the peasants went around saying:" Who shot the El Nino brothers? & # 39 ;, because it was raining every time we wanted it. "
In addition to special weather events such as storms, cyclones and forest fires, accurate forecasts are indispensable for the aviation industry and the $ 60 billion Australian agricultural industry. As Dunford says, "Farmers live by time".
Where do weather observations come from?
The forecasts start from observational data collected on land, in the air and in space: from satellites in polar orbit to airplanes up to balloons that fly up to 15 kilometers. Our weekly forecasts are also international collaborative activities.
Australia is one of 192 member states and territories of the United Nations World Meteorological Organization (WMO) that exchange meteorological observations and data. Most of our satellite images are provided by the Japanese Himawari 8, which orbits Japan, Papua and central Australia.
This sharing extends to meteorological models in which supercomputers – with 1600 billion calculations per second, "super" is not an exaggeration for the Australian version – applying observations of humidity, wind speed, temperature, rain and systems of pressure to the fluid dynamics equations to simulate the evolution of time.
How do supercomputers predict the weather?
The simulations of BoM and CSIRO, as well as those of other regions, which we have used for 60 years, can go beyond seven days – some studies say with precision up to 10 days – but BoM offers seven-day forecasts because that is the threshold to who trust the simulations to be sufficiently accurate.
The "spaghetti model", below, illustrates the divergence of computer models over time. Each color represents a different pressure system and each line is a different "stroke" of a model. If you could imagine that we would have given 10 people a yellow marker with which to draw their predictions, then we compared the level of correspondence with what they had matched, this is what shows the yellow band of lines below (but instead of people with markers , are computers with data). The most orderly image foresees a day in advance and each line has a similar route, which means that the models generally agree and that the meteorologist can be accurate in their predictions.
The most confusing image, above, provides for a week forward and illustrates how the models differ greatly, decreasing the confidence of forecasters in their forecast. In other words, there is less agreement between the different forecasts.
So while we always have a seven-day forecast, behind the scenes a forecaster will be more or less sure of its accuracy depending on the disparity between the models.
Video: a forecaster explains the spaghetti theory
How far can the forecast go?
Seven-day forecasts are issued twice a day, once in the early morning and once in the late afternoon. While forecasts usually become less accurate, McGibbony states that attention is shifting towards a generally accurate forecast.
"Especially a week earlier, people generally want to know how hot it will be and if it will rain. You're not saying there will be rain showers and will produce exactly 5 mm of rain. Let's say c is a percentage chance that there will be rain at a certain point. "
The dott. Andrew Watkins, head of long-range forecasts, says that by the end of this year the office will publish an "average" perspective of two or three weeks to integrate the current three-month seasonal outlook, following the demand from farmers and emergencies Services.
What we will be able to say is that in two weeks there is a high probability that it will be wet or dry, or a hot or cold week.
"It is not so much the weather forecast, we will not tell you in two weeks, Saturday, everything will be fine, c & # 39; is the sun and 23. What we can say is that in two weeks c & # 39 is a high probability will be wet or dry, or a hot or cold week.
"This will help indicate when the time is right to plan crop sowing, or prepare the fire risk plan."
He says this will bring Australia's predictions to a "seamless service, in which you will be able to flow past, present and future".
"In one place you could review the last few weeks, then check the radar for today's rains, see the weekend go well and the low 20s, the next two weeks seem dry, but the winter probably goes on bring the rain, I know I'm a fanatic of the time, but I'd like to see it. "
Improved technology means that forecasting accuracy will invariably increase. One aspect is represented by several observations, for example by satellites that watch cold fronts and cloud bands.
"Fifteen years ago, when I started at the BoM, we received a satellite image every hour," says McGibbony. "Some of the older forecasters here in the 90s said we would have one every six hours. Now we get a satellite image every 10 minutes."
With constantly changing weather systems, they are almost certain to change significantly over a two-week period.
The computing power is increasing, which means that the models can run on higher resolutions in less time and provide more accurate results, according to McGibbony.
But a recent study by researchers at Penn State University in which they tested two advanced meteorological modeling systems found an apparent threshold in which the "chaos theory" takes hold and predictions become conjecture: 14 days.
The study says that with constantly changing weather systems, they are almost certain to change significantly over a two-week period.
While data collection and meteorological modeling improvements will probably make accurate 14-day forecasts in the future, the chaotic and ever-changing nature of weather conditions means much more, for now it is unlikely.
If we have such fantastic computers, what do humans do?
The human role of weather forecasts focuses on the meteorologist who interprets these models, particularly when they diverge significantly. It is their job to evaluate the evidence. The model can be "executed" – started on the computer – many times with slightly different initial conditions.
And in many areas we do not actually have observation equipment, creating data gaps. For example, we do not have a precise measure of wind speed on the ocean, for example 250 kilometers south of Adelaide. But even if a small difference in wind speed would make a slight difference to the model in the next 1-2 days, it could make a significant difference over a week.
Familiarity with Perth's Fremantle Doctor or a "bust of the south" along the NSW coast allows meteorologists to identify them.
"Especially later on, like the seventh day, you could find the most diverse models," says Steven McGibbony, an expert in meteorology at the Bureau of Meteorology.
"It could be suggested that the cold front will come soon, another could say it will be in the afternoon. In this scenario, we look at it and consider which model has worked best in recent times and we weight our forecast towards that model."
The local experience of a nominee also helps: familiarity with Perth's Fremantle Doctor or a "bust of the south" along the NSW coast allows meteorologists to identify them.
"With the arrival of new forecasters, either because they are brand new or have just moved to Melbourne from Darwin, for example, they are addressing people who have been here for a while and say: & # 39 "What do they do normally in this situation?" Says McGibbony, who has worked at the BoM for about 15 years.
"Similarly, if you regularly plan for a particular airport, you may know that if the temperature is below a certain level and the winds are below 10 km / h, there will probably be fog."
Do farmers have their own methods to predict the weather?
"If we could also get 15-day forecasts that were significantly accurate, it would be a great help," says Wayne Dunford, who grew up near Parkes in central NSW for 52 years.
"The most important thing about the forecast is that farmers are spending a significant amount of money to plant a crop in the ground. When you do, there is no humidity and you are relying on in-crop rain events that occur – you are really stretching out to dry. "
With areas of New South Wales and Queensland entering their seventh year of drought, farmers often act based on BoM tools such as three-month rainfall forecasts.
These perspectives have become increasingly accurate since 2013 when BoM switched to a new physics-based computer model that – similarly to the short-term ACCESS model of Australia – uses observations of global oceans, icy atmosphere and earth.
I saw the ants build their nests about six or eight inches higher to avoid water, so we would have had a flood.
But "we are becoming skeptical" after years of "conjecture and false forecasts," says Dunford, who also refers to sources such as the Norwegian website YR, a joint project of the Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation, when planning its harvest sowing.
The dott. Watkins says "chaos theory", in which a very small change can cause a system to behave completely differently much later, inevitably causing long-term climate change.
He estimates that the maximum achievable accuracy of these long-term prospects is 75-80% – in Australia we are sitting 60 to 65%.
"We would need constant observations at every single point on the planet to reach almost 100 percent," he says.
Meanwhile, farmers have their own ways to supplement forecasts.
"I saw the ants build their nests about six or eight inches higher, so as to avoid the water, so we would have had a flood … or in the state forest near us, you wouldn't even know that I'm there, so suddenly you feel the kookaburas coming off and it's raining, "explains Dunford.
If you pass a gidgee tree – also known as a stinking wattle – which is emitting a bad smell, it means that rainfall is close, according to the farmers' belief.
"But you can't rely on those: that's why we rely on BoM to be as precise as possible, even though we know it's complex."
Are the four seasons of Melbourne in a day a myth?
Not as such – but it really applies to the whole south-east of the country.
"One of the unique features of the southeastern Australia," says McGibbony, "is that there is a massive land mass in the north and north-west, and a huge ocean in the south. The north winds they bring warm temperatures from the desert, while the southern winds will bring cool temperatures from the ocean.
"The four seasons in a reputation day come from the fact that you can have very different temperatures."
Why is the forecast sometimes wrong?
When meteorologists in Australia are wrong, it is often in extreme weather conditions and often in summer.
"In the summer you have more energy in the atmosphere with the sun producing more heat, and this is the time of year with more severe thunderstorms," says McGibbony.
"For example, with a cold front, being slightly inaccurate makes a big difference. If you have 40 degrees on one side in front and 20 degrees on the other, a slight disagreement in the models could mean that you're not sure if the heaviest rains they will be in the late morning or early afternoon. "
And if you're a roller coaster who has ever used Perth's climate like a crystal ball, you're ready for something.
"For the southern half of Australia, weather systems will generally follow from west to east. This does not mean that the weather will be the same, because meteorological systems move north and south at the same time and could change, but it is a very general guide. "