Model Development and Enhancement

Cloud microphysical schemes fall into two categories:

1. Explicit or bin schemes that predict mass and number of many different sizes of water drops and ice particles
2. Bulk microphysical parameterizations (BMP) that predict mass/number of total particles of all sizes

Due to the very high computational cost of applying explicit schemes, NCAR–RAL developers focus most of their attention on the BMP schemes for most applications. So–called "single–moment" schemes predict only the mass mixing ratio of the hydrometeors and then diagnose the number concentration by making various assumptions. A "two–moment" scheme also predicts the number concentration and provides more degrees of freedom for representing a size distribution. A recent scheme by Thompson et al, 2008 takes a hybrid approach in order to be computationally efficient while reproducing measurements from various field experiments. The scheme was implemented into the Weather Research and Forecasting (WRF) model and is regularly tested and improved based on results from cloud and precipitation measurements from a variety of convective and stratiform precipitation events.

One of the biggest challenges that NCAR–RAL scientists face when developing these schemes is the intricate and complex interaction between microphysics and other physical processes. For instance, the amount and size of cloud water and ice greatly influence the radiation, which subsequently alters the surface heating and cooling that potentially lead to newly–formed clouds. Likewise, rain falling from convective clouds creates downdrafts that can greatly alter location, strength, and amount of subsequent updrafts and thunderstorms due to assumptions of rain drop or hail size and amount.

Additionally, there are a host of uncertainties within the microphysical schemes in general, some of which are amplified when applied to areas with complex terrain.

• Condensation and "warm" rain processes – Creation of the initial precipitation particles is heavily dependent on the cloud condensation nuclei, which are essentially unknown in day–to–day circumstances. "Autoconversion" schemes that define how droplets collide and coalesce into larger drops are complex and highly variable and often create rain too early or too late compared to observations
• Ice initiation – The freezing of water drops to ice particles is complex and not well understood, partly because it is very difficult to observe and measure in everyday clouds
• Particle size distribution – The water drops and various forms of ice in microphysical schemes assume a simplified mathematical form with regard to numbers of each size. For water, these assumptions dictate when rain begins to form, how rapidly rain descends, and how strong the evaporation rate is below cloud base. For ice, these assumptions greatly affect the resulting growth of ice by vapor deposition and quantity of water vapor and fraction of ice versus liquid
• Particle type issues – There are many uncertainties associated with 1) mass and fall speed versus diameter of ice particles, 2) how to apportion the growth of rime ice between snow and graupel, and 3) the effects of riming on snow particle properties

NCAR–RAL numerical model developers are working on all of these challenging areas in an attempt to find the best optimal solution for each application of the model.