Three-dimensional microphysical and dynamical structures of winter storms in the U.S. Pacific Northwest: 3-year radar climatology and comparisons between operational weather radar observations and regional mesoscale model output

The frequent winter rainfall of the Portland, Oregon region is associated with land-falling cyclones modified by the Coastal Mountain and Cascade Mountain ranges. In the first part of this study, operational WSR-88D radar observations from Portland, OR and upper-air soundings from Salem, OR over three winter seasons (2003-04, 2004-05, and 2005-06) are used to determine a 3D climatology of winter storms. A total of 117 storm events and 2205 hours of radar data were examined. Eighty-four percent of storms have a low level wind direction from the south or southwest, between 158 and 248 deg azimuth. Stability did vary much among storms. Seventy-five percent of storms exhibited values of squared moist Brunt-Vaisala frequency between 0 and 3 x 10-4 s^-2 indicating that most storms were neutral to slightly stable. Low-level wind direction and cross-barrier wind speed are the primary controls on the spatial pattern of precipitation relative to orography. Mean and standard deviation fields of reflectivity, precipitation frequency, and radial velocity are examined as a function of low-level wind direction. In concurrence with findings from MAP and IMPROVE, pre-existing precipitation is enhanced as it passes over the first ridges of the Cascade Mountain range. As expected, the pattern of precipitation is similar among storms with similar low level wind direction. The detailed geographic maps of precipitation frequency can help refine flood forecasts when storms from a particular wind direction are expected.

The second part of the study compares 3D radar observations to MM5 mesoscale model output for the 2005-06 and 2006-07 winter seasons using routine products that can be easily derived from both data sets. We utilize the Penn State/NCAR Mesoscale Model MM5 Version 3.7 in non-hydrostatic mode. A 24-h MM5 simulation is completed twice daily at 0000 and 1200 UTC using 6 hour GFS analyses for initial and boundary conditions. Stationary 1.3 km, 4 km and 12 km nested grids centered on the radar location are nested within a 36 km domain using a 1-way nested interface. For winter 2005-06, model runs utilized the Thompson bulk microphysical scheme, which includes supercooled water and graupel. This scheme was modified in 2006-2007 to include gamma-sized distributions for snow and graupel as well as a Berry autoconversion from cloud water to rain. The mean and standard deviations of radial velocity, precipitation frequency, and mixing ratio are examined for individual storms and storms grouped by prevailing low-level wind direction. Errors in the simulated kinematic and precipitation structures aloft are compared with the model precipitation errors at the surface. Information from the 3D volumes such as vertical profiles and trajectories in the model are used to learn more about the kinematic environment and microphysical processes yielding surface precipitation errors.