JP7J.6 Structural characteristics of winter storms in southern Washington and northern Oregon: Prototypes for routine comparison between regional model output and WSR-88D radar observations

Thursday, 27 October 2005
Alvarado F and Atria (Hotel Albuquerque at Old Town)
Sandra E. Yuter, North Carolina State Univ., Raleigh, NC; and B. A. Colle, C. Spooner, and T. Downing

Major field programs provide comprehensive observations over a few month time period. Once the key phenomena are identified, we can determine the applicability of the new conceptual models using less comprehensive but longer duration observations over a larger sample of storms. Although operational data sets are not equivalent to major field project data sets, judicious use of these data can extend and refine field project case study results. Operational observations can be used to evaluate how well these phenomena are reproduced in real-time numerical model output.

Comparison of model output to observations provides a means of estimating confidence in model output and is useful in diagnosis of errors and evaluation of proposed enhancements. Previous work has shown that comparison of surface fields is necessary but not sufficient. Models can yield plausible surface fields of precipitation with physically implausible 3D precipitation structures, i.e. right answer, wrong reason. 3D fields are needed for comparison to model output. NOAA's WSR-88D radar network provides 3D data on precipitation structure and wind patterns within ~140 km range of the radar. The variable freezing level height of winter storms and wide effective beam width of the radar data limit use of quantitative reflectivities for comparison. Rather, a measure of precipitation frequency based on the reflectivity field is used to compare spatial patterns of precipitation occurrence and persistence. Initial work is focused on the Portland, OR/Vancouver, WA region where winter precipitation has a strong orographic component. As compared to winter storms in the Great Plains, the orographic character of storms near Portland should make it easier for models to predict the spatial distribution of precipitation associated with particular wind patterns. Radar data from the Portland, OR WSR-88D and accompanying MM5 model output are used to generate lowest common denominator products for comparison of storm characteristics. Storm total statistics are used to de-emphasize model timing errors. The wind often veers with height in Portland area winter season storms. Prototype products show that small scale variability in precipitation structures occurs in both horizontal and vertical orientations in response to flow over the small-scale windward ridges of the Cascades.

This study suggests that WSR-88D radar data may be better suited for independent evaluation of regional model output rather than for model initialization. Particularly for regions with orographic forcing, routine (daily) comparison of simple pattern metrics between model-derived and WSR-88D-radar-observed 3D fields could yield valuable information on model performance for a wide range of storm structures and conditions.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner