Characterization of the effect of atmospheric parameters on the infrared sky at the VLT

Knowledge about the external environmental conditions on science data quality is an essential aspect of midinfrared ground-based observations. Before science observations are taken, a standard star must be observed to assess the sky transparency and background, which leads to significant telescope overhead. With data collected by NACO and VISIR instruments in L', M' and N-bands at Paranal, we perform a multivariate correlation analysis between the sky counts and different external conditions (i.e. precipitable water vapour, airmass, humidity and thickness of the dust deposition layer on the main mirror). Using machine learning methods to analyse multiple regression data, we show that knowledge of the external conditions can predict correctly the background sky emission at the relevant wavelengths to within 2-5%. The use of the skycalc tool to verify the predicted background is also briefly described. Our findings have important implications for the operations of the current and future VLT and ELT instruments operating at mid-infrared wavelengths.