How We Know 2014 Was the Hottest Year

Technicians work on a NOAA weather buoy. NOAA

Sure, it’s the middle of winter, and you’re probably more worried about tomorrow’s wind chill than last summer’s heat. But 2014 was particularly warm, as you may have heard—1.24 degrees Fahrenheit higher than the global average for the last century and the hottest year in the history of looking at such stuff. Overall, temperatures have risen 1.4 degrees Fahrenheit since 1880, and, 10 of the warmest years have come after 2002. It’s the sort of trend that should worry you if you hope to avoid rising seas, ferocious storms, droughts, mass extinctions, and other apocalyptic outcomes.

But it also begs a question: Really? Which is to say, how do climate scientists know? What convinces them that they can measure, accurately, the temperature of an entire planet—especially when accurate temperature measurements are crucial for understanding and documenting climate change.

The answer is, it’s not easy. Getting the right data requires a vast network of temperature sensors across all continents and oceans. “It’s a hodgepodge of different networks belonging to a bunch of different owners,” says Deke Arndt, a meteorologist with the National Oceanic and Atmospheric Administration. Here’s how the system works:

Sensors on Land

Some land-basd temperature sensors are housed in a Cotton Region Shelter, which shades the instrument. NOAA

Weather stations, Antarctic research stations, government facilities like water treatment plants, and airports all take regular temperature readings—more than 6,000 land-based sensors, altogether.

The exact types of sensors and the details of how they’re deployed depend on the individual networks that run them, but most rely on thermistor technology—just like the digital thermometer you stick under your tongue. Thermistors are devices whose electrical resistance depends on temperature; electrical current flows more easily when it’s either warmer or cooler, providing a direct measurement for temperature. Many are automated and record data constantly throughout the day, but others require a person to go out and read the measurements daily.

A small fraction of the sensors are old-fashioned liquid and glass thermometers, in which heat causes alcohol or mercury to expand and rise up a calibrated temperature gauge. They may seem quaint, but that old-school tech is actually useful in helping researchers identify biases and better understand the data collected by newer, fancier technologies.

Sensors at Sea

On about 1,500 buoys—a couple-hundred fixed and the rest free-floating—thermistor-based sensors measure the surface temperature of the seas. Some of them sample as often as once a second, transmitting data back home via satellite.

Those buoys are distributed pretty evenly, which means they’re spread way out. Ships cover the gaps, with thermometers hanging off the hull or in the engine room, where they measure the incoming seawater used to cool the engine.

Number Crunching

Collecting the data isn’t enough, of course. Global sensor networks vary depending on location and what organization runs them—like, for example, NOAA’s cooperative observer program, which runs thousands of sensors in North America. The network depends on volunteers who go out every day and read a nearby temperature sensor—one that they may keep in their own backyard. They then report the readings via phone or online. Again, the vast majority of sensors are electronic and thermistor-based, but roughly a quarter are still liquid-and-gas, says NOAA meteorologist Jim Zdrojewski, who helps run the effort. The program started in 1890, and its oldest sensor has been measuring temperatures continuously for 217 years. Now imagine dozens of similar networks, all collecting and sharing data. It’s a challenge.

So algorithms at places like NOAA and NASA crunch through the numbers, figuring out how to take into account increased heat in urban areas and how the spacing between instruments might skew the measurements. The researchers calculate average monthly temperatures and compare them with the averages at every sensor location between 1951 and 1980. These differences, called anomalies, offer a better way to determine trends than simply comparing the temperatures themselves. For example, anomalies accurately track temperature patterns whether you’re looking at water on oceans or cool, mountain air. All in all, it’s not an easy process. But without reliable numbers, no one can tell what the magnitude of the problem is—much less how to fix it.