Errors due to random noise in velocity measurement using incoherent-scatter radar

The random-noise errors involved in measuring the Doppler shift of an ‘incoherent-scatter’ spectrum are predicted theoretically for all values of 1 e /1 i from 1.0 to 3.0. After correction has been made for the effects of convolution during transmission and reception and the additional errors introduced by subtracting the average of the background gates, the rms errors can be expressed by a simple semi-empirical formula. The observed errors are determined from a comparison of simultaneous EISCAT measurements using an identical pulse code on several adjacent frequencies. The plot of observed versus predicted error has a slope of 0.991 and a correlation coefficient of 99.3%. The prediction also agrees well with the mean of the error distribution reported by the standard EISCAT analysis programme.