savannah to pasture, showed regional climate change with a weakened hydrologic cycle exhibiting reductions in both precipitation and evaporation (Lean and Warrilow, 1989). A realistic simulation of precipitation is an important characteristic for studies of climate change. All general circulation models (GCMs) simulate broad features of the observed precipitation pattern, but they also contain significant errors. These include inadequate characterization of the Southeast Asian summer monsoon rainfall and the summer rains in the southern Zaire basin. Recent models also show large differences in estimates of the intensity of tropical ocean rainbelts (Intergovernmental Panel on Climate Change, 1990).
Soil moisture is the "control valve" of the land surface hydrology. Soil moisture is the source of water for evaporation and thus controls heat transfer from the land surface. It also is the principal absorber of heat in the surface. Precipitation and soil moisture, and the associated runoff, are directly interconnected. Soil moisture is an important factor for vegetation, including agricultural crops, and through them affects evapotranspiration, surface reflectivity, and other aspects of the climatic system.
General circulation models appear to be quite sensitive to the proper formulation of the hydrologic budgets of the land surfaces of the earth. For example, numerical experiments reviewed by Mintz (1982) have shown that large-scale changes of land surface evaporation in GCMs produce significant changes in the predicted circulation and precipitation. Smaller and more realistic soil moisture anomalies may not produce such drastic changes, but it appears (e.g., Rowntree and Bolton, 1983) that they can have considerable impact on the climate of the region surrounding the anomally. This and other evidence (e.g., Rind, 1982; Shukla and Mintz, 1982; Sud and Fennessy, 1984; Yeh et al., 1984) indicate that there is a critical need for sound parametric expressions for evaporation and related land surface processes over areas with typical length scales of hundreds of kilometers. While the details of most processes at local scales are well known and understood, as of now there is no agreement on how these hydrologic processes should best be parameterized at the scales appropriate for GCMs. General circulation models are complex in structure, and they involve intense and sophisticated computational schemes. Yet, in most instances their representation of the hydrology of the earth's land surfaces is crude and not well tested.
Climate affects ecosystems in a variety of ways. It is an important influence on processes that determine the carbon and nutrient cycles of