Selection functions of large spectroscopic surveys

Abstract: Context. Large spectroscopic surveys open the way to explore our Galaxy. In order to use the data from these surveys to understand the Galactic stellar population, we need to be sure that stars contained in a survey are a representative subset of the underlying population. Without the selection function taken into account, the results might reflect the properties of the selection function rather than those of the underlying stellar population. Aims. In this work, we introduce a method to estimate the selection function for a given spectroscopic survey. We apply this method to a large sample of public spectroscopic surveys. Methods. We apply a median division binning algorithm to bin observed stars in the colour-magnitude space. This approach produces lower uncertainties and lower biases of the selection function estimate as compared to traditionally used 2D-histograms. We run a set of simulations to verify the method and calibrate the one free parameter it contains. These simulations allow us to test the precision and accuracy of the method. Results. We produce and publish estimated values and uncertainties of selection functions for a large sample of public spectroscopic surveys. We publicly release the code used to produce the selection function estimates. Conclusions. The effect of the selection function on distance modulus and metallicity distributions of stars in surveys is important for surveys with small and largely inhomogeneous spatial coverage. For surveys with contiguous spatial coverage the effect of the selection function is almost negligible.