New ultracool subdwarfs identified in large-scale surveys using Virtual Observatory tools II. SDSS DR7 vs. UKIDSS LAS DR6, SDSS DR7 vs. UKIDSS LAS DR8, SDSS DR9 vs. UKIDSS LAS DR10, and SDSS DR7 vs. 2MASS

Aims. We aim to develop an efficient method to search for late-type subdwarfs (metal-depleted dwarfs with spectral types >= M5) to improve the current statistics. Our objectives are to improve our knowledge of metal-poor low-mass dwarfs, bridge the gap between the late-M and L types, determine their surface density, and understand the impact of metallicity on the stellar and substellar mass function. Methods. We carried out a search cross-matching the Sloan Digital Sky Survey (SDSS) Data Release 7 (DR7) and the Two Micron All Sky Survey (2MASS), and different releases of SDSS and the United Kingdom InfraRed Telescope (UKIRT) Infrared Deep Sky Survey (UKIDSS) using STILTS, Aladin, and Topcat developed as part of the Virtual Observatory tools. We considered different photometric and proper motion criteria for our selection. We identified 29 and 71 late-type subdwarf candidates in each cross-correlation over 8826 and 3679 sq. deg, respectively (2312 sq. deg overlap). We obtained our own low-resolution optical spectra for 71 of our candidates: 26 were observed with the Gran Telescopio de Canarias (GTC; R similar to 350, lambda lambda 5000-10 000 angstrom), six with the Nordic Optical Telescope (NOT; R similar to 450, lambda lambda 5000-10 700 angstrom), and 39 with the Very Large Telescope (VLT; R similar to 350, lambda lambda 6000-11 000 angstrom). We also retrieved spectra for 30 of our candidates from the SDSS spectroscopic database (R similar to 2000 and lambda lambda 3800-9400 angstrom), nine of these 30 candidates with an independent spectrum in our follow-up. We classified 92 candidates based on 101 optical spectra using two methods: spectral indices and comparison with templates of known subdwarfs. Results. We developed an efficient photometric and proper motion search methodology to identify metal-poor Mdwarfs. We confirmed 86% and 94% of the candidates as late-type subdwarfs from the SDSS vs. 2MASS and SDSS vs. UKIDSS cross-matches, respectively. These subdwarfs have spectral types ranging between M5 and L0.5 and SDSS magnitudes in the r = 19.4-23.3 mag range. Our new late-type M discoveries include 49 subdwarfs, 25 extreme subdwarfs, six ultrasubdwarfs, one subdwarf/extreme subdwarf, and two dwarfs/subdwarfs. In addition, we discovered three early-L subdwarfs to add to the current compendium of L-type subdwarfs known to date. We doubled the numbers of cool subdwarfs (11 new from SDSS vs. 2MASS and 50 new from SDSS vs. UKIDSS). We derived a surface density of late-type subdwarfs of 0.040(-0.007)(+ 0.012) per square degree in the SDSS DR7 vs. UKIDSS LAS DR10 cross-match (J = 15.9-18.8 mag) after correcting for incompleteness. The density of Mdwarfs decreases with decreasing metallicity. We also checked the Wide Field Survey Explorer (AllWISE) photometry of known and new subdwarfs and found that mid-infrared colours of M subdwarfs do not appear to differ from their solar-metallicity counterparts of similar spectral types. However, the near-to-mid-infrared colours J-W2 and J-W1 are bluer for lower metallicity dwarfs, results that may be used as a criterion to look for late-type subdwarfs in future searches.