Using EOF Analysis to Identify Important Surface Wind Patterns

Empirical orthogonal functions (EOFs) have been determined for three wind data sets from stations in valleys south of the Great Salt Lake in Utah. Two of the data sets were for summer months, with individual days selected from the MesoWest archive to represent conditions conducive to welldeveloped thermally driven flows. The remaining data set was for the month of October 2000 and was derived from a combination of MesoWest data and data collected during intensive observation periods of the Vertical Transport and Mixing eXperiment (VTMX) conducted in the Salt Lake area. This experiment investigated stable atmospheric conditions in the complex urban terrain around Salt Lake City, Utah. In all three data sets, the primary EOFs represented flows that were directed predominantly along valley axes and were caused by channeled or thermally driven flow. Diurnal variations in EOF intensity showed that thermal effects were the most common causal mechanism. The along-valley EOFs accounted for 43 to 58 percent of the variance in the wind component data sets (8 or 10 stations each). The second EOFs accounted for 13 to 18 percent of the variance. In the summer data sets, the second EOF appeared to represent day-night transition periods; there was evidence of both side canyon flows, and day-night transitional effects in the October data set. The EOF approach has promise for classifying wind patterns and selecting representative cases for simulation or for further detailed analysis.