John Christy and Roy Spencer of the University of Alabama published a series of papers starting about 1990 that implied the troposphere was warming at a much slower rate than the surface temperature record and climate models indicated
Spencer and Christy (1992). One early version of their data even showed a cooling trend
(Christy et al. 1995).
Several groups of scientists began looking closely at this discrepancy. With so many other pieces of evidence indicating warming, it seemed unlikely that the troposphere would not be warming. Errors were discovered in the methods the UAH group used to adjust the data.
To understand what was wrong: The satellites must pass over the same spot on Earth at the same time each day to get a temperature average. In reality the time the satellite passes drifts slightly as the orbit slowly decays. To compensate for this and other orbital changes a series of adjustments must be applied to the data.
The MSU satellite data is collected from a number of satellites orbiting & providing daily coverage of some 80% of the Earth's surface. Each day the orbits shift and 100% coverage is achieved every 3-4 days. The microwave sensors on the satellites do not directly measure temperature, but rather radiation given off by oxygen in the Earth's atmosphere. The intensity of this radiation is directly proportional to the temperature of the air and is therefore used to estimate global temperatures.
There are also differences between the sensors that were onboard each satellite and merging this data to one continuous record is not easily done. It was nearly 13 years after the orginal papers that the adjustments that Christy and Spencer originally applied were found to be incorrect.
Mears et al. (2003) and
Mears et al. (2005).