2 Selection criteria

Our work is part of a large collaboration for optical and infrared studies of radio sources from the WENSS catalogue. For this purpose, three samples of the WENSS catalogue have been defined:

1.

Ultra Steep Spectrum (USS) radio sources with a flux density $100 < S_{\rm 325 MHz} < 500$ mJy and a steep spectral index cut-off of $\alpha_{609}^{325} < -1.1$ ($S \propto \nu^{\alpha}$). Optical and infrared follow-up work has been limited to the 95 per cent of this sample which has no identification on POSS. These objects are powerful tracers of high redshift galaxies.

2.

Gigahertz Peaked Spectrum (GPS) radio sources located in two regions of the survey: one at $15^{\rm h} < {\rm R.A.} < 20^{\rm h}$ and $58^{\rm o} < {\rm dec.} < 75^{\rm o}$, which is called the mini-survey region (Rengelink et al. 1997), centered on the North Ecliptic Pole, and the other at $4^{\rm h}00^{\rm m} < {\rm R.A.} < 8^{\rm h}30^{\rm m}$ and $58^{\rm o} < {\rm dec.} < 75^{\rm o}$. These sources have a spectral index cut-off of $\alpha_{609}^{325} \gt 0.0$ with a convex radio spectrum peaked at a frequency of about 1 GHz (Snellen et al. 1998). They are compact luminous objects, at intermediate and high redshift, which are interesting both as a special class of AGN and as probes of galaxy evolution.

3.

Flat Spectrum (FS) radio sources with an initial selection from the Greenbank Surveys (Condon & Broderick 1985; Gregory & Condon 1991) at 5 GHz ($S_{\rm 5GHz} \gt 20$ mJy). The coordinate limits are $17^{\rm h} < {\rm R.A.} < 19^{\rm h}$ and ${\rm dec}. \gt 55^{\rm o}$ and the spectral index cut-off $\alpha_{609}^{325} \gt -0.5$. They are mostly quasars, up to the highest redshifts, along with a useful number of radio-faint BL Lacs.

The sub-samples of sources to be observed in the IR were defined from the above samples in the following way:

1.

USS sources:

(a)

Sub-sample A with $16^{\rm h} < {\rm R.A.} < 19^{\rm h}15^{\rm m}$ and ${\rm dec}. \gt 55^{\rm o}$ and spectral index $\alpha_{5000}^{327} < -1.2$. At the time of selection optical R-band images and spectra were already available.

(b)

Sub-sample B with $1^{\rm h} < {\rm R.A.} < 2^{\rm h}15^{\rm m}$ and $30^{\rm o} < {\rm dec.} < 40^{\rm o}$. For these objects we had also R-band images and VLA maps.

(c)

Sub-sample C with $23^{\rm h} < {\rm R.A.} < 2^{\rm h}$. It is a high flux control sample supplied by Richard Saunders. It was selected to have a flux density larger than 0.9 Jy at 365 MHz, to be about a factor 5-10 brighter in the radio than the sources in the rest of the sample. For these objects we had only VLA maps.

Given they overlap in right ascension range with the GPS and FS sources, we observed in the IR only 9 out of the 69 objects in sub-sample A. On the other hand, we observed most objects of sub-sample B (20 out of 30) and all 7 objects of sub-sample C.

2.

GPS sources in the mini-survey region further constrained to ${\rm R.A.} \gt 16^{\rm h}$, with optical identification in the R-band. We have imaged in K-band all the 14 objects in this sub-sample. 8 of them were also imaged in the J-band.

3.

No further selection was applied a priori on the FS sample. However given the available observing time, we imaged in K-band 13 out of the 67 sources in the FS sample. These were chosen to cover the full range of flux densities at 5 GHz in the sample and to contain a fair number (5) of sources not identified on deep CCD frames in the R-band.