Skip to Main Content
The Point Spread Function (PSF) is a characteristic of a data acquisition system. In the context of ground-based hyperspectral astronomical observations, it varies both with the observing conditions and as a function of the wavelength. Its knowledge being a prerequisite for many data processing and analysis tasks, it has to be estimated from the data themselves. We propose a simple model to approximate such spectrally varying PSF controlled with only three hyper-parameters. Then we propose to estimate these hyper-parameters from hyperspectral data of an isolated star. The estimation scheme consists of two steps. First, we estimate the star spectrum and the noise variance for each pixel of the datacube; second, we estimate the hyper-parameters. Different estimators are compared for the first step, and accounting for a local average in the wavelength dimension is shown to improve the PSF estimation for low SNR.