Accurate pressure gradient calculations in hydrostatic atmospheric models

Abstract

Most models of atmospheric flow which use the primitive equations require a diagnostic equation to determine local total pressure. In hydrostatic models, this equation is the vertically integrated hydrostatic equation. A frequently used approximation to this integration is to hold the temperature constant within model layers yielding a linear proportionality between δp or δπ (Exner's function) and δz. This procedure yields static pressures with errors on the order of 10⁻³mb.

If terrain following coordinates are used, terms arise in the horizontal momentum equations involving the gradient of total pressure along the coordinate surface, less a correction for the variation of the hydrostatic pressure along a sloped surface. Erroneous horizontal accelerations are common in these models which result from spurious pressure gradients that are due to inaccurate computation of the static pressure. This error may be amplified if the computation of the slope correction term of the horizontal pressure gradient is not consistent with the method of calculating the total pressure.

We derive a methodology to be used in the vertical pressure integrations that is exact if the potential temperature lapse rate is constant between integration limits. The method is applied to both the integration of the hydrostatic equation and the computation of the slope correction term in the horizontal pressure gradient. The method employs a fixed vertical grid and a dynamic one defined by the significant levels in the vertical temperature distribution. With this methodology, the error in calculation of the horizontal pressure gradient acceleration is greatly reduced, especially in situations where the isothermal surfaces are not parallel to the vertical coordinate surfaces. The problem of aliasing and the treatment of significant temperature levels is described.