UGR factor is used for glare analysis of a LAMP; this is an important difference, because for UGR you must consider the dimensions of the output area of the luminous flux of the lamp which may be different by the emitting flux area of the source (a COB led, for example). For this reason you can't find UGR informations about light sources but only about lamps. The projected area can be used for beam analysis and glare analysis with Sollner diagram, but for UGR you must have of all features of a lamp
You can have a look at the CIE publication defining UGR :
CIE 117 (1995), “Discomfort Glare in Interior Lighting”
UGR is not defined for light sources with an angular size less than 0.0003 sr.
For instance, at a distance of 1 m, the light source must be larger than 1.5 cm x 1.5 cm.
Despite this fundamental limitation, lighting manufacturers and designers usually perform UGR calculations on SSL luminaires consisting of multiple small LED sources but incorrectly considering the average luminance over the whole area of the luminaire. This approach is misleading as the resulting UGR is low and does not reflect the physiological perceived glare. Therefore, the use of UGR should be restricted to SSL products with large diffusers, without any visible point sources.
It is more accurate to define the visual comfort by using a luminance ratio criterion. For instance, the French standard on visual ergonomics NF X 35-103 [AFNOR 2013] recommends to limit the ratio of the luminaire luminance to the surrounding luminance to a factor between 20 and 80.
Check the chapter on glare in the following report!
Technical Report Potential Health Issues of Solid-State Lighting
To follow up on Christophe Martinsons's statement that "UGR is not defined for light sources with an angular size less than 0.0003 sr," we have developed a simple model for predicting discomfort glare from outdoor lighting, which usually involves "small" light source. It uses illuminance quantities as these are generally easier to predict using lighting calculation software, and the luminances of backgrounds in real-world scenes are often highly non-uniform. Subsequent work showed that we also had to account for the maximum light source luminance if the source subtended an angle of 0.3 degrees or larger. The model uses the de Boer scale of visual discomfort, a 9 point scale from "just noticeable glare" to "unbearable." A decription of the model can be found online at: http://www.lrc.rpi.edu/programs/solidstate/assist/pdf/AR-DiscomfortGlare.pdf
I also remember that CIE issued a technical report in 2002 defining "UGRsmall" based on luminous intensity, assuming that below a certain angular diameter, intensity is better suited than luminance to assess discomfort glare.