There is general agreement that the global climate warmed between about 1880 and 1940, following several centuries of the "Little Ice Age," which in turn was preceded by the "Medieval Climate Optimum" around 1100 AD. There is less agreement about the causes of this recent warming, but the human component is thought to be quite small. [See BOX] This conclusion seems to be borne out also by the fact that the climate cooled between 1940 and 1975, just as industrial activity grew rapidly after WWII. It has been difficult to reconcile this cooling with the observed increases in greenhouse gases. To account for the discrepancy, the 1996 IPCC Report has focused attention on the previously ignored (direct) cooling effects of sulfate aerosols (from coal burning and other industrial activities), reflecting a portion of incident sunlight. But this explanation to support the "discernible human influence" conclusion is no longer considered as valid. Leading modelers [Tett et al., 1996; Penner et al., 1998; Hansen et al., 1998] all agree that the aerosol forcing is more uncertain than any other feature of the climate models. Models have not yet incorporated the much larger indirect cooling effects of sulfate aerosols (by increasing cloudiness), or the quite different optical effects of carbon soot from industrial and biomass burning and of mineral dust arising from disturbances of the land.
The temperature observations since 1979 are in dispute. On the one hand, surface observations with conventional thermometers show a rise of about 0.1°C per decade, less than half that predicted by most GCMs. On the other hand, satellite data, as well as independent data from balloon-borne radiosondes, show no warming trend between 1979 and 1997 in the lower troposphere, and could even indicate a slight cooling [Christy and Spencer, 1999]. Direct temperature measurements on Greenland ice cores show a cooling trend between 1940 and 1995 [Dahl-Jensen et al., 1998]. It is likely therefore that the surface data are contaminated by the warming effects of "urban heat islands." Some data support this hypothesis [Goodridge, 1996], others do not [Peterson et al., 1999].
While it is certainly true that human life is affected by temperatures at the surface, the GCMs are best validated by observations in the troposphere. It should be noted also that GCMs predict a warming trend that increases with altitude up to about 250 millibars (~12 km), rising to about 0.5°C per decade [Tett et al., 1996] -- in clear disagreement with all observations, whether from the surface, balloons, or satellites.
This is way, way, outdated. Christy and Spencer have been forced -- sometimes willingly, sometimes not so willingly -- to revise the analysis of lower tropospheric temperature data. The trend they have now is about +0.13 C per decade. Other analysis of the same data set finds higher trends. And because the lower troposphere is conclusively warming, this validates the surface temperature record and weighs strongly against any UHI. Possible UHI effects are corrected for in the surface warming data sets, and subsequent studies have found little or even a negative effect. In essence, there is now no significant contradiction between the surface and lower troposphere warming trends.
The following links provide good summaries of the current state of the science on these topics.