Nonstationary Data Analysis

The material presented in previous chapters has been restricted largely to the measurement and analysis of stationary random data, that is, data with statistical properties that are invariant with translations in time (or any other independent variable of the data). The theoretical ideas, error formulas, and processing techniques do not generally apply when the data are nonstationary. Special considerations are required in these cases. Such considerations are the subject of this chapter.


Much of the random data of interest in practice is nonstationary when viewed as a whole. Nevertheless, it is often possible to force the data to be at least piecewise stationary for measurement and analysis purposes. To repeat the example from Section 10.4.1, the pressure fluctuations in the turbulent boundary layer generated by a high-speed aircraft during a typical mission will generally be nonstationary because they depend on the airspeed and altitude, which vary during the mission. However, one can easily fly the aircraft under a specific set of fixed flight conditions so as to produce stationary boundary layer pressures for measurement purposes. The flight conditions can then be changed sequentially to other specific sets of fixed conditions, producing stationary data for measurement purposes until the entire mission environment has been represented in adequate detail by piecewise stationary segments. Such procedures ...

Get Random Data: Analysis and Measurement Procedures, Fourth Edition now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.