Abstract

Laser radar SNR levels in the lower clear atmosphere are high enough to allow consideration of a multiple-watt on-board coherent laser radar, operating in a 3D imaging mode to sense wind shear patterns in the vicinity of airport terminal glide slopes. Any such Doppler sensor must contend with the fact that detection is limited to velocity components parallel to the viewing direction. Hazardous lateral velocities cannot be seen directly and must be inferred by some type of tracking analysis or model recognition procedure. In this paper an analysis is presented which discusses the use of statistical pattern recognition techniques to sense recognizable and recurrent 3D patterns of line-of-sight velocity. The concept is based on the idea that a historical database can be constructed by a combination of measurement, modeling, and simulation that can be used for assessing hazardous flight situations ahead. A Principle Component Analysis of such a database generates a basis set of observed 'eigen features' against which new observations can be compared in a real-time recognition of hazardous patterns. Numerical simulations are used to demonstrate the concepts for a candidate infrared coherent laser radar detection system flying through various combinations of low altitude turbulence, downbursts, and vortex wake environments.

© 1991 Optical Society of America