Microphysics Schemes

Microphysics schemes represent the grid-scale formation and evolution of cloud and precipitation particles: condensation, freezing, collection, riming, melting, and fall. They carry mixing ratios for a set of hydrometeor classes and move mass between them.

The framing that matters is that microphysics handles the resolved condensation, while the cumulus scheme handles the sub-grid kind. That division is why the weight of the microphysics choice depends so heavily on resolution. At 12 km with a convection scheme running, most precipitation comes from the convection scheme and the microphysics matters mainly for stratiform cloud. In a convection-permitting run there is no convection scheme, so all precipitation is microphysical and the scheme’s assumptions set rainfall intensity directly.

That shift is one of the underappreciated consequences of going convection-permitting. Removing one parameterisation increases the influence of another.

What distinguishes schemes

Number of classes. Cloud water and rain at minimum; adding ice, snow, and graupel gives the common six-class configurations such as WSM6. Graupel matters for deep convection, since riming is how convective updrafts produce heavy surface precipitation.

Number of moments. Single-moment schemes predict mass only and fix the particle size distribution’s shape by assumption. Double-moment schemes predict number concentration too, letting the size distribution evolve. Since fall speed depends on particle size, this changes where precipitation lands and how long cloud persists. Double-moment is more physical and more expensive, and its extra freedom is only useful if the additional processes are constrained — which they often are not.

The assumed size distribution in a single-moment scheme is the hidden parameter. It is fixed, it strongly controls fall speed and evaporation rate, and it was chosen from observations of particular cloud types that may not resemble the ones being simulated.

Practical notes

The scheme has to be chosen alongside radiation and convection rather than independently — radiation sees whatever condensate microphysics and convection between them produce, and a mismatch there is the usual source of an unexplained energy-budget bias.

For tropical maritime convection I would want graupel represented, since that is where the intensity comes from. The tropical WRF setup uses WSM6 for the balance between plausible mixed-phase behaviour and runtime.

See also: WSM6, convective rain, cumulus schemes, and the model physics map.