The objective of this effort is to produce explicit predictions of cloud water and ice species for improving forecasts of icing conditions. RAP and MMM are working together toward this goal by using the Penn State/NCAR mesoscale model, MM5. Specifically, the model is being adapted to predict stratiform wintertime clouds which may contain freezing drizzle - a significant aviation hazard.
Currently, the MM5 model contains options of differing microphysical complexities. G. Thompson, K. Manning, R. Rasmussen and J. Coen are exploring methods to forecast freezing drizzle with this model. The model was run in two dimensions for simulated moist flow over a topographic barrier in order to better isolate the microphysics from the dynamics. This design approximates the vertical motions that are typical for observed freezing drizzle cases of isentropic upglide order 10 cm/s.
After completing a control simulation, a series of sensitivity tests were performed. The control run produced unrealistically high amounts of snow and ice and low amounts of cloud liquid water. The sensitivity tests concentrated on the initiation of ice since ice depletes liquid water when the two species coexist. Ice initiation in MM5 occurs via two primary mechanisms: temperature-dependent ice nucleation, and the freezing of cloud drops. Unfortunately, direct or laboratory measurements of these two mechanisms are scarce at relatively warm temperatures (approximately 0 to -12oC) typical of freezing drizzle formation regions. There are a reasonable number of measurements at lower temperatures and, historically, these have been extrapolated to the higher temperatures. There is an increasing amount of evidence which suggests that ice may not initiate very effectively at relatively warm temperatures. Unpublished field evidence provided by A. Cooper, R. Bruintjes, Rasmussen, and P. Schultz suggests that significant ice may not initiate until water supersaturation is achieved. This is a fundamental shift in model parameterizations of the ice initiation process.
Results of the most successful model test run are shown in Figure 1. The left portion of the figure displays temperature in black contours and vertical velocity in blue for upward motion and red for downward motion at 2 hours into the model simulation. The microphysical fields of cloud water and ice are displayed on the right portion of the figure and show an upslope cloud composed mainly of liquid (shown by blue contours) but some ice in the upper portion (shown in red contours). Amounts of liquid water produced by the model are consistent with the magnitude often found in freezing drizzle cases (0.3 g/kg).
Future tests will be performed with adjustments to the ice nuclei activation mechanism and the fall speed of rain. After these tests are finished, we anticipate running the new and improved code on case studies in which freezing drizzle was observed.
