The turbulent heat fluxes near the ground surface are strongly affected by the ability of the surface to redistribute the radiative energy absorbed from the sun and the atmosphere into sensible and latent heat. On a bare dry land, the absorption of this energy results in a relatively strong heating of the surface, which usually generates a strong turbulent sensible heat flux in the atmospheric surface layer and a large soil heat flux. In that case, there is no evaporation (i.e., no latent heat flux) and the Bowen ratio (i.e., the ratio of sensible to latent heat flux) is infinite. By contrast, in wet land, as is common in irrigated agricultural areas and/or after rain events, the incoming radiation is mostly used for evaporation. In that case, the turbulent sensible heat flux and the soil heat flux are usually much smaller than the latent heat flux. As a result, the Bowen ratio is close to zero. When the ground is covered by a dense vegetation, water is extracted mostly from the plant root zone by transpiration. Thus, latent heat flux is dominant even if the soil surface is dry, but as long as there is enough water available in the root zone and plants are not under stress conditions.
Due to these differences in surface fluxes, the characteristics of the atmosphere above dry and wet (vegetated) land are significantly different. The faster heating rate produced in dry land generates a vigorous turbulent mixing and an unstably stratified atmospheric planetary boundary layer (PBL), which typically extends up to a height of 2000--4000 m during the afternoon hours. The slower heating rate of wet (vegetated) land limits the development of the PBL, typically, to a height of less than 1000 m. However, evapotranspiration provides a supply of moisture which significantly increases the amount of water in this shallow PBL. During the afternoon hours, the temperature of the PBL above dry land is considerably warmer than above wet land. During nighttime, however, the strong cooling of bare and vegetated surfaces create almost the same atmospheric inversion [e.g., Avissar, 1992].
Since clouds and precipitation generated in these different PBLs can be quite different, the system of interactions and feedbacks that develop between the land and the atmosphere is very complex. This system needs to account for the movement and storage of water in the ground, the ability of the vegetation to extract the water from the ground, and the dynamical response of the PBL to varying boundary conditions. For instance, Betts et al. [1993, 1994] have shown a close coupling between errors in a model's surface parameterization and its PBL, clouds, and moist convection. An excess of evapotranspiration, by providing excess moisture to the PBL and positive temperature anomalies, acts to destabilize the PBL to moist convection and to promote excess precipitation. Precipitation at one location results in stabilizing subsidence patterns elsewhere. Because of surface heterogeneities, there will occur substantial variations in the instability of PBL air, which may act to reduce the area covered by moist convection.
Thus, in an attempt to provide appropriate lower boundary conditions to the atmosphere, several soil-vegetation-atmosphere transfer schemes (SVATS) have been developed in the past decade. Recent advances on SVATS are provided in Section 2. The Project for Intercomparison of Land-surface Parameterization Schemes (PILPS) is aimed at evaluating how similar and how different these SVATS are [ Henderson-Sellers et al., 1993]. This project, which was recently initiated, is expected to lead to major advances in the representation of land-atmosphere interactions in general circulation models (GCMs). Thus, it is also briefly described in Section 2.
A major assumption adopted in atmospheric models (e.g., GCMs) is that the atmosphere is insensitive to land-surface heterogeneity or, in other words, that the scaling of land-surface fluxes is linear and that a single ``representative'' value can be used to force the atmosphere at its lower boundary [e.g., Sellers et al., 1986; Dickinson et al., 1986]. The validity of this assumption is another major issue which has been investigated during the past four years. Progress in this area is discussed in Section 3.
Finally, Section 4 provides a discussion of the research needs in coming years to improve our understanding and representation of land-atmosphere interactions in GCMs. It is important to emphasize that this paper attempts to summarize the studies that have been conducted in recent years to improve the parameterizations of subgrid-scale land-atmosphere interactions in GCMs. Thus, it concentrates on microscale and mesoscale analyses, and does not discuss the various GCM studies that were conducted to evaluate the impacts and feedbacks of land on the global climate. While these studies provide interesting aspects of land-atmosphere interactions at the global scale (to the extent that the GCM parameterizations correctly represent the investigated processes), they usually do not provide new insights on the development of subgrid-scale parameterizations of land-atmosphere interactions in GCMs, which is the focus of the present paper.