Almost every day, you might check the weather forecast to see what the temperature will be like in the near future or if it might rain.
When you hear a forecast, how exactly do they know what the temperature will be or if it will rain? Also, just how accurate are weather forecasts, and how accurate could they possibly be?
Learn more about weather forecasting, its history, and how it works on this episode of Everything Everywhere Daily.
I’ve done several episodes on the subject of measurement. In particular, I’ve talked about the measurement of time, length, and mass.
All of these are pretty straightforward. Time has natural units like days and years that we use.
Length and mass are just a matter of selecting arbitrary units and using them as a standard.
Temperature is something else entirely.
The only natural point that humans can really use as a reference point is the freezing point of water at standard pressure.
Beyond that, there is no intuitive way to measure temperature. Without a special tool, you can’t really quantify what the temperature is, unlike length, mass, or time.
Understanding how weather works took a long time to develop. For the longest time, all we knew about the weather was what we could experience from our immediate surroundings.
The first attempts at formal weather prediction began with the ancient Babylonians. They attempted to predict the weather based on cloud patterns and astrology, which made their predictions more akin to clairvoyance than science.
The Greek philosopher Aristotle wrote a book called Meteorologica, where he set out his theories of how the weather and the natural world worked.
Unfortunately, pretty much everything he said was wrong. For example, he didn’t believe that the wind was moving air.
The problem wasn’t so much that Aristotle was wrong. He was writing over 2,000 years ago when there was no real systematic way of doing science.
The problem was that the words of Aristotle were treated as fact in the West for centuries. Aristotle was used to end inquiry or conversations just because he was assumed to be right.
In China, India, and the Islamic world, they too, had systems for making short-term forecasts, mostly by observing the clouds in the morning and evening.
It wasn’t until the invention of the thermometer that it was possible to start treating meteorology as a real science.
In the late 16th century, Galileo created a device known as a thermoscope, which is today known as a Galileo thermometer. It consists of various glass balls of differing masses inside a sealed cylinder filled with water. They will rise and sink depending on the temperature.
This could detect changes in temperature, but there was no numeric scale to measure.
In 1612, the Italian Santorio Santorio created the first device with numeric measurements, which was horribly inaccurate.
The first modern thermometer with a standardized scale was created in 1714 by a man whose name you might recognize, Daniel Gabriel Fahrenheit.
Although it might seem like the Fahrenheit scale makes no sense with 32 and 212 degrees for the freezing and boiling point of water, but there actually was a method to his madness.
One hundred degrees was supposed to be the temperature of the human body, although it was later discovered to be a bit less than that.
The difference between 32 and 212 is 180, which was by design, as he thought that freezing and boiling were 180 degrees apart from each other, just like opposite points on a circle.
Fahrenheit was a glassblower who also created some of the best mercury and alcohol thermometers in the world at the time.
Another closely related invention that also provided weather data was the barometer.
The Italian scientist Evangelista Torricelli was the first to realize that Aristotle was wrong and that air and the atmosphere had weight. He developed the first barometer, which could measure atmospheric pressure.
One final device was the development of the anemometer, which could measure wind speeds. The first conceptual description of such a device was in the 15th century by Leon Battista Alberti. However, the first practical one was developed in 1846 by the Irish physicist John Thomas Romney Robinson. It was a four-cup spinning device that measured the speed of the wind.
With the thermometer, anemometer, and barometer, it was now possible to start collecting actual weather data. However, there was still something missing to create weather forecasts.
Gathering temperature and pressure data was a huge step forward, but you could only do it for one palace. What was needed was a way to rapidly exchange weather information with other places. Once you had that information, you could start to make predictions.
The technology which made that possible was the telegraph. Before that, the rate that information could travel was slower than the weather.
The first attempts to systematically track and forecast weather were made by the Royal Navy. In particular, Rear Admiral Francis Beaufort and Vice-Admiral Robert FitzRoy.
The wreck of the steamship Royal Charter in 1859 due to a storm off the coast of Wales led FitzRoy to develop a storm warning service.
Taking daily weather data from 15 stations via telegraph, the Royal Navy began publishing their first storm warnings. In fact, it was Robert FitzRoy who coined the term “weather forecasting.”
Fun Fact, FitzRoy was the Captain of the HMS Beagle, which took Charles Darwin to the Galapagos Islands.
Daily weather forecasts first appeared in the London Times newspaper in 1861. That year also saw the publication of the first weather maps.
As the telegraph spread, more data was able to be collected. In 1870, the United States created the National Weather Service. They began collecting data on November 1st with 24 recording stations. It was originally part of the Department of War but was transferred to civilian control in 1891.
Early forecasts were not precise and were often vaguely worded. For example, an 1870 forecast simply read, “Through tomorrow, probably fair with a fresh breeze.”
The next big advance in understanding the atmosphere was the use of balloons. French meteorologist Léon Teisserenc de Bort pioneered the use of balloons to study the upper atmosphere. His studies led to the discovery of the tropopause and stratosphere.
The development of numerical weather predictions began in the 1920s. English mathematician Lewis Fry Richardson attempted to numerically develop a six-hour forecast for two points in central Europe. The time it took to calculate everything was six weeks.
Richardson proved, at least in theory, that accurate mathematical predictions could be made using mathematics and the laws of physics, although there were some massive problems with his prediction based on poor initial data.
The problem was that technology in the 1920s wasn’t good enough to calculate a forecast in the time the forecast was to take palace.
It wasn’t until the 1950s that computers were able to be used to solve the problem.
Here I should note one of the biggest problems with predicting the weather. Weather is what is known as a chaotic system. That means that small changes in the initial state can lead to large changes down the road.
This is best exemplified in what is known as the Butterfly Effect. A butterfly that flaps its wings in the rainforest could possibly lead to a hurricane later on.
Weather is so chaotic that even if we had data collection points situated one meter apart from each other across the entire surface of the Earth, going up 10,000 feet, and all the possible computing power, the best we could hope to do was have a somewhat accurate forecast for 30 days out. Beyond that, it becomes impossible.
In 1955 the US began publishing the first regular numeric weather forecast. This is simply a forecast where actual temperatures and rainfall amounts are given. This was soon followed over the next twenty years by many other countries.
All numeric forecasts are based on models of the atmosphere. Overtime, as more data has been collected and computers has gotten better, the models have become better and better.
The UK recently installed the world’s most powerful weather forecasting computer that cost £1.2 billion pounds. It can make predictions down to one square kilometer for the entire country, and as small as 300 square meters for certain critical areas such as airports.
As computers were geting more powerful, so was data collection.
In 1959, the United Sates launched the world’s first weather satellite, Vanguard 2. It only flew for 19 days, but it was able to provide images of cloud cover over the parts of the Earth where it was daylight.
Today, all weather satellites are either in polar orbit, where they can cover the entire Earth once per day, or they are in geostationary orbit, where they can observe one half of the Earth permanently.
Satellites can now not just record cloud cover, but track storms, record temperatures in the upper atmosphere, measure water temperatures and wave heights, and the extent of snowpack.
One of the most important developments in weather forecasting has been the use of weather radar.
Weather radar was discovered by accident during World War II when radar operators noticed that rain or snow would often cause echoes.
After the war, several researches pursued trying to finding a use for these radar echos. These efforts led to the ability to measure the average size of raindrops and reflectivity, which could be used to determine the amount and intensity of rainfall.
One of the first televised uses of weather radar occurred in 1961 when a young local Texas reporter by the name of Dan Rather went to a weather radar facility in Galveston, Texas to cover Hurricane Carla.
He got permission to broadcast live from the site and got the managers of the facility to draw a rough outline of the Gulf of Mexio on a transparent sheet of plastic. He then held the sheet over the radar display to show viewers the size and location of the storm.
Thanks to his efforts in displaying the scope of the storm to viewers, hundreds of thousands fled, and there were only 35 deaths, compared to the 12,000 dead from a similar hurricane in 1900.
Today, doppler radar is the norm for weather radar, which can not only detect position and intensity, but also velocity.
In the United States, especially in the midwest, many television stations compete on the basis of their proprietary weather radar.
Today, there are weather networks that share the data they collect. The thing about the weather is that it is more accurate for everyone the more data which is shared.
There are over 1000 sites around the world which launch weather balloons once or twice a day.
Likewise, many weather radar systems are part of networks where they share their data to create comprehensive radar maps.
The global group which oversees the sharing of weather data is the World Meteorological Organization, which is an agency of the United Nations.
There are even more attempts underway to gather even better data.
One is to get individuals to network personal weather stations. Personal weather stations are smaller and cheaper than professional ones, but when they combine their data on the internet it can be quite powerful.
WeatherUnderground is a network with over 250,000 personal weather stations around the world.
One future development for data gathering on the horizon is weather drones. Right now, weather balloons are single-use devices. They go up and fall back to Earth at some unknown location.
A weather drone could be launched and fly on a predetermined route, at a fixed altitude, and then fly back to be reused. They would be more expensive than weather balloons, but they could be reused and carry more and better data gathering equipment.
Weather drones still aren’t very common, however, they are starting to see use. The National Oceanic and Atmospheric Administration flew a drone into Hurricane Ian in 2022 for the first time to get measurements inside the eye of the storm as well as on the eye wall.
There are still issues being worked out with weather drones because they have to fly in conditions that a military drone might not have to. This includes having deicing systems in the drone for flying in cold temperatures or very high altitudes.
Over the last few decades, weather forecasting has gotten better and better. Five-day forecasts are accurate about 90 percent of the time, seven day forecast are accurate about 80 percent of the time, and ten-day forecasts are accurate about 50 percent of the time.
There are some meteorologists who have launched smartphone apps which claim to make predictions down to the minute up to one hour in advance, however, there is a great deal of skepticism about such claims.
Weather forcesating is big business. Transportation, agriculture, tourism, and many other industries are heavily reliant on accurate forecasts. Forecasting itself is a $4.5 billion dollar business, with hundreds of billions more depending on it.
The next time you hear a weather forecast, think about how much that goes into it. It involves recording stations, balloons, satellites, and radar. The data from which is all constantly being churned through super computers.
It has come a long way from ancient Babylonians trying to decipher the clouds.