Estimating the spectral resolution of a radio data cube from its noise properties

Abstract

We present a new method to estimate the spectral resolution of a radio data cube from its noise properties. We assume that the data cube is the result of the convolution of a spectrally-uncorrelated cube with a constant convolution kernel K. We demonstrate that K and the resolution, defined as K's integral, can be estimated thanks to the analytical relation between the autocorrelation of the noise spectra extracted from the cube and the power spectrum of the kernel. We discuss a number of algori…