Improving Cloud Simulation in WRF Through Assimilation of GOES Observations

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 8 January 2015: 4:30 PM
131AB (Phoenix Convention Center - West and North Buildings)
Andrew T. White, University of Alabama, Huntsville, AL; and A. P. Biazar, R. T. McNider, B. Dornblaser, and K. Doty

Clouds directly modulate the radiation budget over their area of influence. Changes in the cloud cover, therefore, can have pronounced effects on the meteorological conditions of a given area. Unfortunately, numerical meteorological models fall short of accurately producing clouds at the correct time and location with respect to observations. This results in an inaccurate representation of the atmospheric state throughout the model domain. These errors inhibit the model's ability to accurately predict variables such as temperature and radiation, and lead to a misrepresentation of vertical mixing and inaccurate development of the boundary layer. Through assimilation of Geostationary Operational Environmental Satellite (GOES) cloud observations within the Weather Research and Forecasting (WRF) model, cloud placement in time and space within the model can be improved. Using GOES cloud observations, it is possible to determine where a model overpredicts and underpredicts cloud cover. Knowing these disagreement areas between the model and observations, a method has been developed to analytically solve for vertical velocities necessary to produce or dissipate clouds within the model system. By introducing positive vertical motion in areas where the model underpredicts cloud coverage, air parcels can be lifted to the point of saturation, creating a cloud. Likewise, the introduction of negative vertical motion in areas that the model overpredicts cloud coverage allows air parcels to sink and warm to a point where they are no longer saturated. This method was tested for the August 2006 time period. Preliminary results of this method show an increase in the cloud agreement between the model and GOES observations by a daily average of greater than six percent over the tested time period. Hourly cloud agreement improvement approached ten percent. Results from the use of this method will be presented.