Distinguishing Between CAT and Non-CAT Areas by Use of Discriminant Function Analysis

Abstract

The formulation of algebraic functions, involving synoptic-scale atmospheric parameters as variables, capable of predicting clear-air turbulence within 7000 ft sub-layers of the stratosphere was attempted. The data sample used was composed of 153 turbulent and non-turbulent regions identified from 46 stratospheric flights of the XB-70 aircraft over the western United States during the period March 1965 to November 1967, and the values of 69 synoptic-scale parameters determined from rawinsonde data associated with each of the regions. After the XB-10 regions and the values of the synoptic-scale parameters were grouped into one or more of five overlapping categories, or sub-layers, determined by the altitude of the aircraft at the time the turbulence or non-turbulence was reported, discriminant function analysis was employed in each sub-layer to construct functions which could discriminate the turbulent from the non-turbulent regions. Those discriminant functions yielding the best results in each sub-layer were tested from 23 stratospheric flights of the YF-12A aircraft over the same area during the period March 1970 to January 1972. For each sub-layer, five discriminant functions yielding the best results were used to derive a forecasting procedure. This procedure correctly identified approximately 85% of the turbulent and non-turbulent regions in each of the five sub-layers.