Abstract: It is shown that the Middleton’s Class A model can be made to fit the amplitude probability distribution (APD) of the extremely low frequency (ELF) atmospheric noise measurement. For the accurate APD of ELF noise, the measured noise is processed with an FFT filter to suppress the 50 Hz interference. The non-Gaussian property of ELF noise is testified by Lilliefors hypothesis testing. By considering the APD and the Quantile-Quantile plot of real ELF atmospheric noise based on a parameter estimation of Class A noise model, excellent agreement between the measured data and the theoretical curves provided by the model can be obtained.

Key words:

non-Gaussian noise, amplitude probability distribution, class A noise model