Identification and study of white dwarfs in J-PLUS Data Release 1

Abstract

Context. The Javalambre Photometric Local Universe Survey, with 12 wide and narrow photometric bands, enable the reconstruction of spectral information that can be key to characterize the astronomical sources observed in the survey. Aims. Making use of the data published in the first data release we build a method to identify white dwarfs using photometry and parallaxes, to obtain a catalogue of WDs candidates. Methods. The classification is made using the random forest classifier, a machine learning algorithm. The first step is to select available sources to build training and test datasets. For this we use the available white dwarfs catalogue using information from Gaia DR2 and select the rest of objects (non-WDs) using sources outside the white dwarfs locus in the color-magnitude diagram. Different attempts of the classification model are done and discussed, until finding the best option incorporating progressively the information from the 12 J-PLUS passbands, 3 Gaia passbands, parallaxes from Gaia DR2 and their errors. Results. Parallax is the most important information for the classification of WDs, as well as the extreme bands, the bluest (u, J0378, J0395) and the reddest (i, J0861, z). The resulting candidates are well placed in many diagrams. The final catalogue contains 24634 white dwarf candidates present in J-PLUS first data release, with information from 15 different filters (12 from J-PLUS and 3 from Gaia), parallax and a probability evaluating how reliable the classification is. Conclusions. The classification model developed is be a very promising tool for the identification of white dwarfs using J-PLUS data, which will be improved in future data releases using data for all the northern celestial hemisphere.