A recent comparison (1) of temperature readings from two major climate monitoring systems - microwave sounding units on satellites and thermometers suspended below helium balloons - found a "remarkable" level of agreement between the two.
To verify the accuracy of temperature data collected by microwave sounding units, John Christy compared temperature readings recorded by "radiosonde" thermometers to temperatures reported by the satellites as they orbited over the balloon launch sites.

He found a 97 percent correlation over the 16-year period of the study. The overall composite temperature trends at those sites agreed to within 0.03 degrees Celsius (about 0.054° Fahrenheit) per decade. The same results were found when considering only stations in the polar or arctic regions.

"The idea was to determine the reliability of the satellite data by comparing it to an established independent measurement," Christy said. "If satellite data are reliable when the satellites are over the radiosonde sites, that means you should be able to trust them everywhere else."

The 99 radiosondes reported an aggregate warming trend of 0.155 degrees Celsius (about 0.28° Fahrenheit) per decade since 1979. Over those 99 spots on the globe, the satellites also recorded a warming trend: 0.128 degrees Celsius (about 0.23° Fahrenheit) per decade.

Globally, however, the satellite data show a cooling trend of 0.03 degrees Celsius per decade since the first NOAA TIROS-N satellites went into service.

"These 99 radiosonde launch sites are just not distributed evenly around the planet," Christy said. "They are not representative of the total globe."

Radiosonde balloons are released from stations around the world, usually at noon and midnight Greenwich standard time. As each balloon climbs from the surface to the stratosphere, the temperature is measured and relayed to the ground by radio.
While there are more than 1,000 radiosonde launch sites globally, the data from many sites either are not readily available or are not consistently collected. Christy used data from 99 sites at which there has been long-term systematic and reliable data collection. These 99 radiosonde launch sites are in a box bounded by Iceland, Trinidad, Truk Island and Alaska.
In an earlier study, an upper air temperature record compiled by NOAA from 63 daily weather balloon sites around the world indicated a 17-year climate trend of -0.05° C per decade, which was in exact agreement with the satellite data at that time, Christy said.


With nine satellites measuring the temperature over periods of from one to six years, a method was devised to merge all the data into a single, consistent time series.

Each satellite has its own bias that, if not calculated and removed, would introduce spurious trends. The biases are calculated by directly comparing each satellite with others in operation at that time. Periods of overlapping operation ranged from three months to three years, and were sufficient to determine these biases.

Because the MSU instruments are so stable and have so many thousands of observations, the biases between the satellites are known to within 0.01 deg. The final product removes these biases so that all data are referenced to a common base. (2)
To check the final product, comparisons were made over a 16-year period with balloon measurements as stated above, and the phenomenal agreement provided the independent validation necessary to conclude that the merging technique developed for this dataset was accurate.


Of great concern to scientists is the lack of consistency in the way readings are taken and in the thermometer surroundings. Since most thermometers for which long-term records exist are in towns and cities, the effects of population growth and the construction of nearby roads, parking lots, runways and buildings may cause the temperature to rise a little due of urbanization. This temperature change may be an artifact of a local "asphalt effect" rather than a long-term widespread climate change.


While the temperature data collected by ships at sea is reported as a sea surface temperature, this data reflects water temperatures from about three to 60 feet below the surface - the level from which water is drawn into the ships.
The thousands of individual thermometers used to collect this data are not calibrated against a scientific standard, nor is there a method for verifying the accuracy of either the thermometers or the reports matching temperature readings to specific times and places.

Only in places where there are many overlapping observations can there be any confidence in their accuracy.


  • 1. J.R. Christy, 1995, "Climatic Change," Vol. 31, pp. 455-474.
  • 2. J.R. Christy, R.W. Spencer and R.T. McNider, 1995, "Journal of Climate," Vol. 8, pp. 888-896.
  • 3. R.W. Spencer, J.R. Christy and N.C. Grody, 1990, "Journal of Climate," Vol. 3, pp. 111-1128.
  • 4. R.W. Spencer and J.R. Christy, 1992, "Journal of Climate," Vol. 5, pp. 858- 866.
  • References