The debates about satellite MSU data and climate models: the role of `simple' empiricism

The long-running debates concerning the satellite-based MSU readings of tropospheric temperature trends, especially those in the tropics, involved, among other things, the treatment of radiosonde data as basic or foundational. John Christy, Roy Spencer, and co-authors frequently appealed to the radiosonde data as an independent source of confirming evidence for their interpretation of the satellite raw data, the UAH dataset. This type of ‘simple’ empiricism clashed with the approach used by other data analysts and modelers to the radiosonde datasets, which can be best characterized as ‘complex’ empiricism. While a straightforward and simple approach to the radiosonde data, wherein the data are seen as transparently reflecting the atmospheric temperature and temperature trends, may be appropriate for some uses, the majority of data analysts and modelers in this debate found such data to be embedded in too many assumptions and with too much uncertainty to be treated in a plain or simple way. Their complex empiricist approach emphasized that the radiosonde data were highly dependent on the choice of station, time of day, and condition and details of the instrument, among other factors. Thus, the radiosonde datasets could not be taken as straightforwardly representing the ‘real’ temperature trends, and thereby could not serve as an authoritative independent test for confirming the satellite temperature trends. The complex empiricist approach, under which theory, models, and data are all used in concert to produce scientific outcomes, resulted in an appeal to, and the development of, alternate means of independently testing the satellite datasets, as well as a drive to uncover and remove the remaining biases in both the radiosonde and satellite datasets. This approach was essential to the resolution of this long-standing debate in climate science.