Global temperatures are compiled for various areas including global (land-ocean), global (meteorological stations), three latitude bands,
and hemispheric, by the Goddard Institute for Space Studies of NASA.
The graph shows the annul mean global surface temperature anomaly from the base period 1951-1980, for the period 1880 - 2019.
For details see data.giss.nasa.
The basic GISS temperature analysis scheme was defined in the late 1970s by James Hansen when a method of estimating global temperature change was needed for comparison with one-dimensional global climate models. The scheme was based on the finding that the correlation of temperature change was reasonably strong for stations separated by up to 1200 km, especially at middle and high latitudes. This fact proved sufficient to obtain useful estimates for global mean temperature changes.
Temperature analyses were carried out prior to 1980, notably those of Murray Mitchell, but most covered only 20-90°N latitudes. Our first published results (Hansen et al. 1981) showed that, contrary to impressions from northern latitudes, global cooling after 1940 was small, and there was net global warming of about 0.4°C between the 1880s and 1970s. The early analysis scheme went through a series of enhancements.
The analysis method was fully documented by Hansen and Lebedeff in 1987 including quantitative estimates of the error in annual and 5-year mean temperature change. This was done by sampling at station locations a spatially complete data set of a long run of a global climate model, which was shown to have realistic spatial and temporal variability. A more complete uncertainty analysis was published in Lenssen et al. in 2019).
As there are other potential sources of error, such as urban warming near meteorological stations, many other methods have been used to verify the approximate magnitude of inferred global warming. These methods include inference of surface temperature change from vertical temperature profiles in the ground (bore holes) at many sites around the world, rate of glacier retreat at many locations, and studies by several groups of the effect of urban and other local human influences on the global temperature record. All of these yield consistent estimates of the approximate magnitude of global warming.
Further affirmation of the reality of the warming is its spatial distribution, which has largest values at locations remote from any local human influence, with a global pattern consistent with that expected for response to global climate forcings (larger in the Northern Hemisphere than the Southern Hemisphere, larger at high latitudes than low latitudes, larger over land than over ocean).
An updated documentation by Hansen et al. in 2010 compares alternative analyses and addresses questions about perception and reality of global warming; various choices for the ocean data are tested; it is also shown that global temperature change is sensitive to estimated temperature change in polar regions, where observations are limited. A multi-year smoothing is applied to fully remove the annual cycle and improve information content in temperature graphs. Despite large year-to-year fluctuations associated with the El Niño-La Niña cycle of tropical ocean temperature, the conclusion could be made that global temperature continue to rise in the 21st century.
For further information on a wide range of weather/climate matters.. Google: climatediceandthebutterfly