Determination of Meso-Gamma Scale Wind Flow in Complex Terrain Utilizing Python
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 5 January 2015
Identification of representative wind flow patterns can be used as a forecasting tool in complex terrain environments. Preliminary analysis was conducted for hourly data from 13 meteorological stations located at the Nevada National Security Site (NNSS) from December 2013 through April 2014. A two-step process was utilized to find the dominate flow patterns, along with secondary and tertiary sub-classes within each class, using Python's “scikit-learn” module, which is built on NumPy, SciPy, and matplotlib. The first step conducted a rotated Principal Component Analysis on the zonal and meridional components of the wind. Through this process, five dominate wind patterns are identified which explain approximately 80% of the variance in the dataset. The correlation matrix revealed the meridional wind variability is similar in all areas of the site, whereas zonal wind variability primarily accounts for the differences between each wind flow pattern. In the second step, a k-means cluster analysis was performed on the combined meridional and zonal components of the wind from each of the five leading rotated principal components, illustrating the primary flow for each respective principal component.
Preliminary results indicate the meteorological stations within complex valleys tend to present more variability than stations outside the influence of terrain. The most dominant wind pattern is representative of a typical nighttime flow, in which the surface wind decouples from the synoptically-driven flow, resulting in winds moving slowly from high to low elevation. The second dominant wind pattern is typical of the flow after passage of a cold front in which strong north to northwest winds prevail.
Future analysis will extend the dataset through September 2014 with an emphasis to be placed on characterizing the primary flow patterns for each season.