ABSTRACT
This chapter introduces a complementary function, called the spectral density function, which is the natural tool for considering the frequency properties of a time series. Some statisticians initially have difficulty in understanding the frequency approach, but the advantages of frequency methods are widely appreciated in such fields as electrical engineering, geophysics and meteorology. The power spectral distribution function describes how the power is distributed with respect to frequency. In the case of a time series, the variance may be regarded as the total power. The adjective ‘power’, which is sometimes prefixed to ‘spectral distribution function’, derives from the engineer's use of the word in connection with the passage of an electric current through a resistance. For a sinusoidal input, the power is directly proportional to the squared amplitude of the oscillation.
