isparity of temperature trends (1979–99) of atmosphere and surface: can we trust the surface data?

A National Research Council report [1] has highlighted the disparity between surface temperature trends (from thermometer readings) versus those from MSU instruments in weather satellites and radiosondes carried in weather balloons. The global surface record shows a substantial warming trend in the last twenty years while the atmospheric data from both MSU and radiosondes show little if any warming. As is well understood, climate models predict the opposite, namely a stronger atmospheric warming trend than for the surface.

We have calculated linear temperature trends (in degree C per decade) in order to establish the influence of the unusually warm El Nino year of 1998. Our OLS coefficients for the surface (GHCN v2), radiosonde (NCEP Reanalysis), and MSU data (Channel 2 Lower Troposphere) are as follows:

	'79-'95	'79-'96	'79-'97	'79-'98	'79-'99
Surface			+0.080	+0.109
Rad-sonde (500mb)			-0.017	+0.033
MSU	-0.006	-0.003	+0.002	+0.046	+0.036

As can be seen, the MSU trends are close to zero before 1998. Its trend value for 1979-1999, with 1998 deleted, is -0.0003. In order to better understand the disparity between surface and atmospheric data, we investigated the latitude, altitude (for NCEP), and seasonal dependence of all the trends. We also calculated separate trend values for two periods, '79-'98 and '89-'99.

We conclude from this analysis that the surface trends are not credible. Further evidence comes from the fact that the well-controlled surface temperature data for the United States do not show an appreciable post-1940 warming, in contrast to the global surface data. In addition, proxy data from tree rings, ice cores, etc. show no post-1940 warming: Many even show a cooling trend after 1940.

1. National Research Council. "Reconciling Observations of Global Temperature Change." National Academy Press. Washington, DC. Jan, 2000.