temperature cycle, although the modeled diurnal range in mid-latitudes is generally less than the observed.

Cao et al. (1992) also conducted a number of experiments with a one-dimensional radiative-convective model (RCM) to show that the decrease in the diurnal temperature range with doubled CO₂ in that model is due primarily to a water-vapor feedback. A reduction of only 0.05°C in the DTR was observed when the absolute humidity was held constant. Table 5 indicates that increased sensible-heat exchange and evaporation are also important factors leading to a reduction in the DTR in RCM simulations with enhanced CO₂.

Interestingly, the ratio of the DTR decrease relative to the increase of the mean temperature is closer in the RCM to the observed ratio over the past several decades than it is in the GCM (Table 5). The RCM omits the positive feedbacks to the DTR from reductions in cloud and surface albedo included in the GCM simulations (Cao et al., 1992). These feedbacks tend to increase the DTR because of reduced atmospheric (cloud cover) and surface (snow cover) albedo. Other GCMs simulate both increases and decreases in cloudiness from global warming—decreases in much of the troposphere, but increases in the high troposphere, low stratosphere, and near the surface in high latitudes (Schlesinger and Mitchell, 1985). A tendency for a general increase in cloud cover over land (which now seems likely from the observational evidence) could help explain the large discrepancy between the observed data and the model projections of the ratios of the decrease in the DTR range relative to the mean temperature increase. (This explanation assumes, of course, that the recent warming is induced by increases in anthropogenic greenhouse gases. On the other hand, it leaves questions regarding the cause of the apparent change in cloudiness and how it has affected the mean temperature.)

The ability of present-day GCMs to adequately simulate changes in the DTR resulting from enhanced CO₂ is also affected by surface parameterizations of continental-scale evaporation. As Milly (1992) points out, present-day GCMs can overestimate the surface evaporation because of their failure to properly account for the cooling that occurs with the evaporation. Milly raises concerns about the veracity of the results from studies of soil-moisture changes induced by an increase of greenhouse gases. Accurate projections of the changes in the surface boundary-layer DTR with increases of anthropogenic greenhouse gases will be strongly dependent on adequate simulation of these processes.

Given the dependency of the DTR on surface-layer processes, interactions with the land surface, and cloudiness (which are all areas of significant uncertainties within pres-