A computationally efficient linear semi-lagrangian scheme for the advection of microphysical variables in cloud-resolving models

E. Gavze, E. Ilotoviz, A. Khain*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

2 Scopus citations

Abstract

An efficient semi-Lagrangian (SL) numerical scheme for the advection of positive-definite microphysical variables in cloud-resolving models is proposed. The scheme is a linear combination of SL schemes of the first and second order. The proposed scheme is compared with two high-order, monotonic, nonlinear Eulerian schemes, in two 2D tests. In the first test, advection of a positive scalar field with large gradients is performed within complicated idealized steering flows. In the second test, a hail storm is simulated using the 2D Hebrew University Cloud Model, describing the interactions between 15 size distributions of different microphysical quantities (each distribution is described by 43 mass bins). The proposed scheme is computationally efficient, has a low numerical diffusion and a high level ofmass conservation accuracy, and preserves the sum of multiple advected variables as well as the linear relationships between the advected scalar variables. Although the proposed SL scheme is not exactly positive definite, the negative values and their number, which may appear only in the case of very strong gradients, are negligible. The scheme produces results similar to those of nonlinear Eulerian advection schemes while reducing computation time by approximately a factor of 10.

Original languageAmerican English
Pages (from-to)3427-3452
Number of pages26
JournalMonthly Weather Review
Volume148
Issue number8
DOIs
StatePublished - Aug 2020

Bibliographical note

Publisher Copyright:
© 2020 American Meteorological Society.

Fingerprint

Dive into the research topics of 'A computationally efficient linear semi-lagrangian scheme for the advection of microphysical variables in cloud-resolving models'. Together they form a unique fingerprint.

Cite this