The risk tropospheric ozone poses to forests in the United States is dependent on the variation in ozone exposure across the distribution of the forests in question and the various environmental and climate factors predominant in the region. All these factors have a spatial nature, and consequently an approach to characterization oi ozone risk is presented that places ozone exposure-response functions for species as seedlings and model-simulated tree and stand responses in a spatial context using a geographical information systems (GIS). The GIS is used to aggregate factors considered important in a risk characterization, including: (1) estimated ozone exposures over forested regions, (2) measures of ozone effects on species' and stand growth, and (3) spatially distributed environmental, genetic, and exposure influences on species' response to ozone. The GIS-based risk characterization provides an estimation oi the extent and magnitude of the potential ozone impact on forests. A preliminary risk characterization demonstrating this approach considered only the eastern United States and only the limited empirical data quantifying the effect oi ozone exposures on forest tree species as seedlings. The area-weighted response of the annual seedling biomass loss formed the basis for a sensitivity ranking: sensitive-aspen and black cherry (14%-33% biomass loss over 50% of their distribution); moderately sensitive-tulip popular, loblolly pine, eastern white pine, and sugar maple (5%-13% biomass loss); insensitive-Virginia pine and red maple (0%-1% loss). In the future, the GIS-based risk characterization will include process-based model simulations of the three- to 5-year growth response of individual species as large trees with relevant environmental interactions and model simulated response of mixed stands. The interactive nature of GIS provides a tool to explore consequences of the range of climate conditions across a species' distribution, forest management practices, changing ozone precursors, regulatory control strategies, and other factors influencing the spatial distribution of ozone over time as more information becomes available.

The onset of convection in shear flow driven by lateral heating and also uniformly heated from below is investigated numerically by Galerkin's method. Stress-free as well as rigid, perfectly conducting boundaries are considered. The analysis is valid for small and moderate Prandtl numbers. The magnitude of the lateral basic temperature gradient may be expressed by a dimensionless Grashof number G, while the uniform heating from below is represented by a Rayleigh number Ra. Depending on the values of G, Ra and the Prandtl number Pr, a variety of interesting situations arise. In particular it is demonstrated that the form of the most unstable mode, i.e. whether it is a roll with axis aligned along the basic flow (a longitudinal roll) or one with axis normal to the basic flow (a transverse roll), depends on the value of the Prandtl number. For small values of G, the marginally stable disturbances are found to be steady, while for larger values of G, oscillatory instability occurs. For all values of G considered here (G [lsim ] 3000), computations of the energy balance for the marginally stable disturbances show that the main instability mechanism is of thermal origin, while the effect of shear may be important in selecting the preferred mode of disturbance.

The flows induced by the presence of an insulating sloping boundary in a doublediffusive system are examined. In the diffusive case, when the component with the larger diffusivity is unstably distributed, it is known that under certain circumstances horizontal motions are induced near the slope, and that a series of horizontal layers forms. We investigate the formation and properties of the layers, in particular their vertical scale and its dependence on the stratification and the slope angle. The scale of the layers is found to be a strong function of Gρ, the ratio of the vertical density gradient of the unstably distributed component to that of the stably distributed component. At low values of Gρ, no layering was observed; at larger values of Gρ layers were formed, and their scale increased as Gρ → 1. A weak dependence of scale on slope angle was also observed with the scale diminishing as the angle of the slope to the horizontal increased.A new form of layering has been observed when the basic stratification is in the finger sense. At high enough values of Gρ the basic stratification is unstable to finger motions and these exist throughout the fluid. When a slope is introduced, horizontal motions are set up near the slope which cause the fingers to break down and layers are produced. There is considerable horizontal motion in these layers as well as convective motions driven by the fingers in the interfaces between the layers. The formation of these layers and some of their properties are documented.