Three of the world’s most complete temperature tracking records – from NASA’s Goddard Institute for Space Studies, the National Oceanic and Atmospheric Administration’s National Climactic Data Center and the UK Meteorological Office’s Hadley Centre – begin in 1880. Prior to 1880, temperature measurements were made with instruments like thermometers. The oldest continuous temperature record is the Central England Temperature Data Series, which began in 1659, and the Hadley Centre has some measurements beginning in 1850, but there are too few data before 1880 for scientists to estimate average temperatures for the entire planet. Data from earlier years is reconstructed from proxy records like tree rings, pollen counts and ice cores. Because these are different kinds of data, scientists generally don’t put proxy-based estimates on the same charts as the “instrumental record.”
The above-mentioned agencies and others collect temperature data from thousands of weather stations worldwide, including over the ocean, in Antarctica and from satellites. However, instruments are not perfectly distributed around the globe, and some measurement sites have been deforested or urbanized since 1880, affecting temperatures nearby. Each agency uses algorithms to filter the effects of these changes out of the temperature record and interpolate where data is sparse, like over the vast southern ocean, when calculating global averages. Generally, all three datasets agree quite closely (see graph above) and are in agreement on the trend of global warming since the Industrial Revolution.