Cloud Formation and Explicit Precipitation

Physical Processes

Cloud Formation

Clouds are formed as air parcels are forced to rise, cool and condense. This forcing can be accomplished in several ways including surface heating, frontal lifting, or mixing of the air (fog). In warm clouds, droplets can grow by condensation in a supersaturated environment and by colliding and coalescing with other cloud droplets. Cloud droplets can also form with the aid of cloud condensation nuclei in an unsaturated environment (RH > 90%). If a cloud extends above the 0°C line, it is called a cold cloud. Even though the temperature may be below 0°C, water droplets can still exist in clouds as super-cooled droplets. As a matter of fact, cloud temperatures frequently need to get below -10°C for any significant number of ice particles to form. If a cold cloud contains both ice particles and super-cooled droplets, it is a mixed cloud. If a cold cloud consists entirely of ice, it is said to be glaciated. Since cloud droplets are small, they are hardest to freeze. Condensation nuclei help the cloud droplets to freeze.

Precipitation Formation

In warm clouds, growth by condensation slows as the droplet radius increases. There must be another means for warm clouds to produce precipitation. Warm cloud droplets grow and form precipitation by the collision-coalescence process. In cold clouds, precipitation forms by deposition, riming, and aggregation. The growth of ice crystals, first by deposition from the vapor phase in mixed clouds, and then by riming and/or aggregation can produce precipitation-sized particles in reasonable periods of time (see figure below). Precipitation sized particles actually start falling once gravity overcomes the upward vertical movements of the drops in the cloud sustaining them.

Model Implementations

Grid box Clouds

ETA Model

Temperature > 0°C cloud is liquid.
Temperature < -15°C cloud is ice.
Temperature between 0°C and 15°C cloud is ice if ice cloud existed in box or box above during previous time step.
There is only one phase of cloud water computed per grid box.
Ice clouds form snow and liquid clouds form rain.
Clouds condense when relative humidity is > 75% for land and > 80% for ocean.

Rapid Update Cycle (RUC)

The RUC uses the NCAR/Penn State mesoscale model MM5 parameterization for cloud formation. Cloud hydrometeor species represented:

Weather Research and Forecast (WRF)

To be determined

Nested Grid Model (NGM)

Cloud cover is not computed.

NOGAPS

Clouds are not explicitly predicted.

Spectral Clouds

Global Spectral Model (GSM) and Regional Spectral Model (RSM)

Clouds can exist in most model layers except near the surface and above the tropopause.
Stratiform clouds are computed from mean cloud and relative humidity relationships.

AVN/MRF

Cloud parameterization scheme predicts cloud liquid and ice based on RH and temperatures in each model layer.
Clouds can be advected and partial cloudiness is allowed.

Grid box Precipitation

ETA Model

A 3-D array indicates if precipitation is solid or liquid.
Precipitation within a box can be liquid and solid.
A snow water equivalent of 10" snow = 1" water is used because snow depths not reported.
Two sources of condensation are large scale and convective processes.
According to Mike Staudenmaier, Jr. in "The Explicit Cloud Prediction Schemein the Meso-Eta Model," these are the 6 microphysical processes are represented the ETA model:

Sublimation, deposition, evaporation, condensation, melting, and freezing of raindrops is considered.

RUC

Included in the model is a freezing level for precipitation type determination.
Model removes supersaturation as precipitation.
Recognizes orographically induced precipitation.
Evaporation of precipitation occurs with convective precipitation but not with stable precipitation.
Convective precipitation is formed on a sub-grid scale.
Stable precipitation is formed on a grid scale.
The RUC uses the NCAR/Penn State mesoscale model MM5 parameterization for precipitation formation. Precipitation species represented:

NGM

Grid scale precipitation occurs when relative humidity > 95%.
Relative humidity < 95% causes evaporation of precipitation.
All precipitation falls as rain.
Snow depth analysis is not available.

Weather Research and Forecast (WRF)

To be determined

NOGAPS

Precipitation is not explicitly predicted.

Spectral Precipitation

GSM and RSM

Rain evaporates in unsaturated layers in the model below the condensation level.
Precipitation falls as snow if temperatures in the lowest model layer and at the ground are less than 0°C.
The grid scale condensation scheme: excess moisture is precipitated into lower layers resulting in ground precipitation.
Arakawa-Schubert cumulus parameterization scheme gives smaller resolution in the precipitation scheme.

AVN/MRF

Diagnoses rain and/or snow rates based on the amount of cloud liquid or ice predicted.

Advantages and Disadvantages

Advantages

GSM and RSM can have precipitation that falls as snow.
ETA considers sublimation, deposition, evaporation, condensation, melting, and freezing of raindrops.
ETA is simple enough to be used in operational meteorology, in terms of computational speed and storage.
ETA and RUC both show cloud droplets that are super-cooled. This is important for aviation but does not matter for weather forecasting.
ETA and RUC recognize orographically induced precipitation.
RUC recognizes land/ocean differences in precipitation formation.
AVN/MRF does a good job of predicting precipitation generated by large cyclonic storms, due to well predicted large scale motions.

Disadvantages

The NGM does not compute cloud cover.
There is no stratiform rain in the NGM.
All precipitation from the NGM falls as rain. This would not be good in regions of the world where frozen precipitation is common.
NGM and ETA do not report snow depth. This would again not be good in regions of the world where frozen precipitation is common.
ETA does not consider precipitation advection.
ETA is not initialized with cloud information.
AVN/MRF does a poor job at predicting lake effect or orographic snow events.
AVN/MRF cannot predict isolated convective storms because the model does not forecast for convective motion

Conclusions

The RSM is a local version of the GSM.
The RUC parameterizations are very similar to those used in the ETA model. The RUC can be used as a short range ETA model.
The RUC seems to be the 'best' model to use for cloud cover and explicit precipitation as it uses cloud data when initialized and uses an almost identical scheme for explicit precip formation .
AVN is a good product for synoptic scale predictions in the field due to its updating of initial conditions four times a day.
MRF initially is good for the beginning of its forecast period of synoptic events, but lacks credibility compared to AVN as initial conditions are updated.

References

Benjamin, Stan, NOAA, 3 January 1998: "Information on RUC/MAPS."
http://maps.fsl.noaa.gov/MAPS.40km.html.

Benjamin, Stan, NOAA/ERL Forecast Systems Lab, 16 June 1994: "Implementation of the RUC."; http://maps.fsl.noaa.gov/tpbruc.cgi.