An evaluation of size-resolved cloud microphysics scheme numerics for use with radar observations. Part 2. Water vapor diffusion

 
Poster PDF

Authors

Hyunho Lee — NASA - Goddard Institute for Space Studies
Ann M. Fridlind — NASA - Goddard Institute for Space Studies
Andrew Ackerman — NASA - Goddard Institute for Space Studies

Category

Microphysics (cloud, aerosol and/or precipitation)

Description

In the study of the evolution and characteristics of cloud drop size distributions (DSDs) and the formation of drizzle, a size-resolved (bin) microphysics scheme can be beneficial because it describes DSDs without any predefined functional form and is able to simulate the evolution of arbitrarily shaped DSDs. However, schemes used to calculate each process included in a bin microphysics scheme are prone to numerical diffusion especially at the tails of DSDs. Use of cloud radar observations to constrain simulations places additional demands on scheme numerics at the large-drop tail. Here we extend our previous study on the numerics of the stochastic collection equation to those of the water vapor diffusion equation. Three methods used to solve the water vapor diffusion equation in a one-moment bin microphysics scheme are examined: 1) a method that distributes drops with arbitrary mass into two adjacent bins to conserve the number and mass of drops (“2mom”), 2) a method similar to 2mom that conserves three moments using three adjacent bins (“3mom”), and 3) a piecewise parabolic method that treats the condensation and evaporation as advection along the mass axis (“PPM”). A near-analytic solution is obtained using a dynamic mass bin grid and provided as the reference solution. Results using a simple box model reveal that all the methods converge, albeit somewhat slowly. Among the examined methods, the PPM converges most rapidly. In addition, all the methods converge faster on an arithmetic grid than on a geometric grid. When combining condensation with collection, relatively weak numerical diffusion while solving condensation causes relatively distinct differences on solving collection, particularly in terms of the moments of radar Doppler spectra. Large-eddy simulations that simulate weakly drizzling stratocumulus are performed using the aforementioned three methods and Doppler spectra obtained using a forward simulator are compared with observations.