The greenhouse effect predicts that adding carbon dioxide should warm Earth. Many independent measurements show that it has. Scientists trust the conclusion because so many different records point the same way.
- Air temperature: averaged over the globe, the decade 2011 to 2020 was about 1.1 °C warmer than 1850 to 1900. Each of the last four decades was warmer than any decade before it since 1850 (IPCC).
- Carbon dioxide: about 280 parts per million before the 1800s; above 420 ppm in recent years, measured at Mauna Loa, Hawaii (NOAA).
- Sea level: rose about 20 cm between 1901 and 2018, as ocean water warmed and expanded and land ice melted (IPCC).
- Ice: glaciers are shrinking around the world, and Arctic sea ice in late summer covers far less area than when satellite records began in 1979.
Key idea
Thermometer records, rising seas, shrinking ice and the rising carbon dioxide record all show that Earth has warmed over the last century, by about 1.1 °C. The warming shows in averages over decades, not in any single year.
Why it works
Single years bounce up and down with weather patterns such as El Niño, so one year can be cooler than the year before. Averages over 10 or 30 years smooth out the bounces and reveal the trend. The scientific assessment (IPCC, 2021) concludes it is unequivocal that human activities, mainly burning fossil fuels, have warmed the atmosphere, ocean and land.
Worked example: Rate of rise
Carbon dioxide at Mauna Loa was about 340 ppm in 1980 and about 420 ppm in 2023. What was the average rise per year?
- Change: $420 - 340 = 80$ ppm.
- Time: $2023 - 1980 = 43$ years.
- $80 \div 43 \approx 1.9$ ppm per year.
Answer: About 1.9 ppm per year
Worked example: How much more?
By what percent has carbon dioxide risen from 280 ppm to 420 ppm?
- Increase: $420 - 280 = 140$ ppm.
- Percent: $140 \div 280 = 0.5 = 50\%$.
Answer: About 50%
Common mistake
Judging the trend from one hot or cold year. Years bounce up and down; the warming shows in averages over decades. Check: compare the average of one decade with the average of another, not one year with the next.