Abstract: Reflectivity estimates from clouds and other clutter targets normally reqe an averaging of a large number of independent radar pulse samples. If the radar generates noise pulses of a predetermined bandwidth and pulse length, a large number of returns can be derived and, when averaged over the full length of the pulse, the returns provide the desired reflectivity estimate. Considering a noise source whose spectrum in the frequency domain can be represented by a rectangular function of width .beta. (bandwidth), the properties of this function are such that the time (.tau.) at which it first becomes zero is related to .beta. by .tau.=1/.beta.. Time .tau. is the correlation time. If the pulse length T is made relatively long compared to the correlation time, it becomes possible to average or integrate within the pulse length T since the number of uncorrelated noise configurations is T/.tau.=.beta.T. For a noise pulse of one microsecond long (T) and a bandwidth (.beta.