Effects of Upstream-Biased Third-Order Space Correction Terms on Multidimensional Crowley Advection Schemes

Robert E. Schlesinger Department of Meteorology, University of Wisconsin, Madison, WI 53706

Search for other papers by Robert E. Schlesinger in
Current site
Google Scholar
PubMed
Close
Restricted access

Abstract

Several authors have improved the stability or conservation accuracy of the highly popular second-order Crowley advective differencing scheme in multidimensional flow. However, these improved versions still have the marked dispersive phase lag of the original combined or time-split Crowley scheme. This leads to poor shape preservation of a numerically advected disturbance, and anomalous oscillations upstream of it.

This paper presents a method for reducing the dispersive phase lag in the Dukowicz–Ramshaw and Smolarkiewicz versions of the combined multidimensional Crowley scheme, here designated as Types I and II respectively. The method involves adding upstream-biased corrections for third-order spatial truncation error, analogously termed Types I and II. This approach preserves the one-step explicit nature of the schemes.

The effects of the third-order space correction terms on stability and phase error are evaluated in two dimensions, using linear Fourier component analyses and numerical experiments that advect a cone-shaped scalar distribution in steady nondivergent flows.

Our main findings are: (i) our Type I correction dramatically improves the phase accuracy of the Type I scheme, with only slightly less amplitude preservation or stability; (ii) the Type I scheme also performs much better with our Type I correction than with the Hill third-order correction, which is three times as large with excessive damping and large dispersive phase lead; (iii) despite a larger stability region than its Type I counterpart as emphasized by Smolarkiewicz, the Type II scheme has greater dispersive phase lag and faster damping; (iv) applying the Type I correction to the Type II scheme provides wider stability but much less effective phase error reduction than for the Type I scheme; (v) the phase accuracy of the Type II scheme is dramatically improved by the Type II corrector, but with a highly restrictive stability criterion.

Abstract

Several authors have improved the stability or conservation accuracy of the highly popular second-order Crowley advective differencing scheme in multidimensional flow. However, these improved versions still have the marked dispersive phase lag of the original combined or time-split Crowley scheme. This leads to poor shape preservation of a numerically advected disturbance, and anomalous oscillations upstream of it.

This paper presents a method for reducing the dispersive phase lag in the Dukowicz–Ramshaw and Smolarkiewicz versions of the combined multidimensional Crowley scheme, here designated as Types I and II respectively. The method involves adding upstream-biased corrections for third-order spatial truncation error, analogously termed Types I and II. This approach preserves the one-step explicit nature of the schemes.

The effects of the third-order space correction terms on stability and phase error are evaluated in two dimensions, using linear Fourier component analyses and numerical experiments that advect a cone-shaped scalar distribution in steady nondivergent flows.

Our main findings are: (i) our Type I correction dramatically improves the phase accuracy of the Type I scheme, with only slightly less amplitude preservation or stability; (ii) the Type I scheme also performs much better with our Type I correction than with the Hill third-order correction, which is three times as large with excessive damping and large dispersive phase lead; (iii) despite a larger stability region than its Type I counterpart as emphasized by Smolarkiewicz, the Type II scheme has greater dispersive phase lag and faster damping; (iv) applying the Type I correction to the Type II scheme provides wider stability but much less effective phase error reduction than for the Type I scheme; (v) the phase accuracy of the Type II scheme is dramatically improved by the Type II corrector, but with a highly restrictive stability criterion.

Save