Previous observational studies of wintertime frozen precipitation events have suggested that episodes of high pollution, in the form of increased nucleating aerosols, can effectively reduce frozen precipitation efficiency by reduced riming. Higher concentrations of activated aerosols increase cloud droplet concentration while reducing the mean droplet size. Reduced droplet size results in decreased collection efficiencies for ice species collecting cloud droplets. Thus, the ice species contain less mass and experience less gravitational settling to the surface.

The Regional Atmospheric Modeling System (RAMS) at Colorado State University was recently outfitted with a new cloud droplet parameterization that nucleates cloud droplets from activation of CCN and GCCN within a lifted parcel. The activation of CCN and GCCN and growth of their solution droplets results in direct formation of small and large cloud droplets with respective diameter ranges from 2-40 and 40-80µm. This method replaces the former treatment, whereby, all cloud droplets nucleate with an initial diameter of 2µm if supersaturated conditions exist. The previous method leaves no room for initial formation of larger droplets that could realistically begin formation from activation of large natural or anthropogenic aerosols. The new scheme gives the model flexibility in terms of initial aerosol concentrations and sizes and the tracking of activated and unactivated aerosols throughout cloud lifecycles.

The chosen simulation event is a winter snowfall episode over Colorado that occurred from 28-29 February 2004; this case represents a classic, northwesterly flow, high mountains snowfall event for Colorado and provides an excellent test case for examining the sensitivity of the model microphysics to the initial aerosol concentration. It further provides an objective comparison to the daily-run realtime RAMS forecast model to assess any forecast improvement brought about by updating the cloud droplet parameterization.

Simulations were initialized with varying concentrations of CCN (from 50 - 1000 /cm3) and GCCN (from 10-5 - 10-2 /cm3) so as to assess the sensitivity of the underlying model microphysics to changes in initial aerosol concentrations. Among the sensitivity experiments, the total domain-averaged accumulated precipitation varies by a maximum range of ~1.0 mm (~10% variability). At the Storm Peak Lab location the total precipitation variability reached 5mm (~20%) among simulations. The range of precipitation extremes over six SNOTEL sites northern Colorado reveal a variation up to ~30% depending upon the model initialization of aerosol concentrations.

The total accumulated precipitation was minimized when using the standard droplet parameterization since this scheme essentially behaves as if the air is extremely polluted. Generally, greater precipitation accumulation and variability, along with better agreement to most observations sites, occurred with use of the aerosol parameterization. The greatest variability in specific hydrometeor-type accumulation exists for graupel since it is the model microphysical species that is the heaviest rimer of cloud droplets. The rimed mass of small cloud droplets was greatest for the lowest CCN and GCCN concentrations. The riming of large cloud droplets was also greatest for the lowest CCN concentrations, but was greater for higher GCCN concentrations since these result in direct nucleation of large droplets with high collection efficiency.